luigi circuit ng 1:08.354 (rta strats tas) Sir Corvid 2026-09-17 | luigi circuit ng 1:08.354 (rta strats tas)
[MKWii TAS] DS Yoshi Falls (No Glitch) - Automatic - 0:57.572 Corvid 2025-07-26 | BKT cut: 69 ms - ♪ Chris Christodoulou - Once in a LullabyAutomatic claims a second NG track! Many thanks to @Alego11-TAS for making this possible, and @MarthTAS for previous help on the shroom. Check out Alego's "making of" video here: youtube.com/watch?v=rlDViaBtwjICompared to manual transmission, auto is unable to hop or charge MTs, but it can turn without speed loss. Auto tends to lose time everywhere due to these limitations, but its turning ability gave it a niche on Peach Gardens, where manual bikes have to do many slow realignments. On Yoshi Falls, auto saves a lot of time on the bridge, and has boost panels to accelerate after drifts, but the waterfall just seemed to kill any hope of viability for auto... That was not enough to stop us.I race against two ghosts, which are the 641 and 768 shown at the beginning. Here are the improvements over the 768:- Much better movement at first boost panel. Somehow it was missed before.- A new discovery, called A-tech. By releasing A at the end of a boost, we don't suffer any speed loss and the bike's rotation is affected weirdly. Out of a drift, this lets us turn tighter in the direction of the drift. We implemented this before the waterfall, and at the last turn. (It's not useful for the shroom.)- The waterfall was improved by using the manual strat; get big airtime by touching the soft walls at the edge of the track. Previously we only got 26-28 f air, which landed earlier than the auto method, but now we were able to get 31 f air which ends up clearing a bit more waterfall. I'm not 100% sure but I think A-tech made the difference, it keeps the bike pointed a bit more upwards.
[MKWii TAS] N64 Mario Raceway (No Glitch) - 1:38.895 Corvid 2025-07-17 | BKT cut: 100 ms - 𝅘𝅥𝅮 Falling in Reverse - Game Over (Nightcore)HUGE improvement, we're collecting a lot of coins with this one! @EjayB found a new strat for the shroom, we optimized it, then I resynced the rest of the run following previous input by @abbymkw12 and @sealpuptakingalittlenap.Racing is a game, and we have been looking for a better way to it. This new strat saves 40 ms; however, the timesave only amounts to 33.3 ms (2 frames) each lap. This is because both pipe tricks (especially the second one) are harshly QM locked, and the previous run had just the right QM for optimal tricks without any time loss. Oh well, take a rest, push the pause, level up and move along.The improvement consists of getting a jammed trick at the ramp, then wheelie off the ramp while the trick effect is stored. Getting the jammed trick is hard, it was a big obstacle but we're jumping hurdles here. Although the airtime is much larger than before, the previous strat wasted most of the MT boost by triggering the trick boost so early. Not only that, the large airtime allows for a bounce landing and slipchain; this lets us eliminate the chain before next turn. Truly a big boy strat! Another one on Ejay's tally, for keeping score.
[MKWii TAS] SNES Ghost Valley 2 (No Glitch) - 0:51.791 Corvid 2025-06-26 | BKT cut: 665 ms - 𝅘𝅥𝅮 King Gizzard and the Lizard Wizard - Crumbling CastleNew TAS on this track after over 10 years! Thank you to everyone who helped get this done: @Alego11-TAS, @Aoiro_mkw, @campbellmop355, @abbymkw12, @EjayB, @sealpuptakingalittlenap (and thanks @Paul-hq7gf for adding the water, very important)Do not be deceived by the short length or flatness, this track is tough to TAS. The biggest challenge was the blocks, which have super wonky collision; no matter if we drift tighter or wider, it usually counts as a hit anyway. The shroom is also annoying, even one input change can make a massive difference.Few details of the run I find interesting:- The late slipdrift at the start is about a frame faster. It is unbelievably precise. I don't know how ego and sorry even discovered the inputs.- The last two turns are the biggest QM locks, due to the soft wall interactions. On laps 2 and 3, we have to waste a few ms to get the 88 u/f drift before the boost panel, otherwise we'd always drop to 81 u/f.- On lap 3, the falling blocks are almost a pacelock. The GV2 blocks always cause the bike to slow down and bounce away from them, so on the third turn we have to get a small bounce to clear over the falling blocks. Had we been faster, the blocks may have been potentially impossible to drift over.- Side trick is ~5 ms faster at the end, but we have to use up/down trick on laps 1 and 2. This is due to side tricks causing tilt lock, which makes the shroom much slower.
Mushroom Gorge Spear TAS - Making of Corvid 2025-06-13 | Watch the full run here! youtube.com/watch?v=bbF8zl-H7sYI worked in private with @Alego11-TAS and @Monster3rum to make an updated MG run without ODM. We ended up finding a lot of gradual optimizations, particularly in the cave section. So I made this video to document the TASing process and show off what worked and what didn't. Kind of a weird video but I hope it's enjoyable!The biggest "potential" improvements for the future are:- 98 u/f mushroom abuse at the end of the cave. This requires a very specific pattern of bounces when landing on the mushroom. We tried a lot with all sorts of approaches, but it just doesn't seem to be possible. Not doing mushroom skip for a better setup also seems to lose too much to be worth it.- Improving the start of lap 1 by a frame. Despite Monster and I's best efforts, we only got ~10 ms and the turn after the mushroom requires very specific bounces for a low hop, which we couldn't replicate.
[MKWii TAS] Moo Moo Meadows No-Glitch - 1:13.274 Corvid 2025-05-28 | BKT cut: 195 ms -- 𝅘𝅥𝅮 Pokémon Scarlet/Violet - East Province (Riding)Updated TAS for the cow track, just in time for the cow kart game! Thank you to all the people who contributed inputs: @Alego11-TAS, @Blounard, @abbymkw12, @Firefanged, @Kierio04, @MKWiiTomasThe previous TAS aged well for being 6 years old, not many improvements were found since. The main time save is performing the blue ramp lowtrick without dropping the wheelie, saving ~30 ms per lap.We also saved ~20 ms on the first turn on laps 2 and 3; it doesn't look clean, but it's the best we can do with the Spear, and much better than holding the MT for longer.There are other smaller improvements throughout, particularly the startslide and the last turn.Miscellaneous notes:- The moles leave a trail of offroad behind them, which fades away after a while. On lap 2, we are now too fast for the offroad trail to disappear before the boost ramp. This cost us ~33 ms, unfortunately there is no way to mitigate this. The ramp approach is also slightly different, due to another mole trail; down trick worked better than left trick for whatever reason.- On lap 3, the cows just barely get in the way. Luckily, we can dodge them by simply reworking the line through the previous turn.- The blue ramp is the most notable QM lock. Only a narrow range of positions for starting the RFH give the fastest low. The shroom is slightly improvable on each lap, but the timesave cannot be carried.- The slipchain on the first turn of laps 2 and 3 helps us fix the bad rotation from the aggressive first drift. The optimal airtime (3 frames) also relies on strict QM.
[MKWii TAS] N64 Sherbet Land No-Glitch - 2:00.492 Corvid 2025-04-26 | BKT cut: 285 ms -- 𝅘𝅥𝅮 Chris Christodoulou - Risk of Rain 2After so many setbacks, this track was finally conquered. With this TAS, the total No-Glitch time drops below 54 minutes (53:59.818), which will be the last minute barrier for NG 3lap.Thank you @Alego11-TAS for helping me with the cave and shroom on every lap, and to @EjayB and @abbymkw12 for helping to optimize the last turn of lap 3.Also thanks to @xMeap826 for the awesome texture mod I used! (from Nostalgic Texture & Music Pack)Here's a rough breakdown of the improvements from the TAS we published last year:- ~100 ms by using Baby Daisy instead of Toadette. Toadette has a small speed and large offroad bonus; Baby Daisy has a large speed and small mini-turbo bonus. Baby Daisy drives ~0.55 u/f slower than Toadette on ice, but ~0.6 u/f faster in the cave, and about 65% of the track is ice. However, the MT bonus gives Baby Daisy 4 more frames of boost per MT, which really adds up over time.- ~130 ms by getting airtime at the end of the cave for a slipdrift. @EjayB found out that it's (barely) possible to get a very late 2 f bounce, which leads into the next turn nicely.- ~50 ms from better optimization, particularly at the first and last turns. These are tough to get right, and with this being my third time TASing rSL now, I am a lot more experienced.A few more notes:- For RTA runs, Baby Daisy is much closer to Toadette, because 3 MTs fewer are done every lap. In theory, she should still be ~10 ms faster per lap, but this calculation does not factor in Toadette's offroad traction bonus (helps with drifting on ice). Baby Daisy can certainly get WR, but whether she is actually faster than Toadette is very unclear to me.- The penguin on lap 3 is very close to being a pacelock. We have to drive wider to dodge it, costing 4 ms due to the shape of the cave. With more timesaves earlier it will likely become impossible to dodge.- The main QM lock is airtime after the 4th MT in the cave. The start of laps 1 and 2 is slightly improvable, but then we would need to lose time to get good QM in the cave.
[MKW TAS] GBA Shy Guy Beach (No Glitch) - 1:19.868 Corvid 2024-06-23 | BKT cut: 307 ms (by Luke, DwainiumB)𝅘𝅥𝅮 Alestorm - Call of the WavesSpecial thanks to @Blounard for creating Advanced Ghost Comparison. This TAS, along with his recent work on rDKM NG, is the first to make heavy use of this tool. AGC uses position data rather than inputs, so we can delay or advance the ghost as much as we want... even entire laps! This allowed us to very efficiently progress after lap 1, and recycle inputs after backtracking entire laps. Without Blounard's tool, this run would have taken way longer with a worse finishing time.This also sets a new record for most authors for a single run, from 12 to 15. Who would've thought a no-glitch TAS of a flat track would be the biggest collab TAS of all time? Thank you @abbymkw12 for assisting me with optimization through most of the course; @Citrinitaser for handling the encode; @mkwLuke and @MKWiiTomas for early strat testing; @campbellmop355 for backtracking the TAS back to lap 1; and @EjayB, @ellucinea, @Jellopuff_TAS, @jtlonchannel, @Kierio04, @MarthTAS, @PyraminxstarsHD, @SirLucas., @sealpuptakingalittlenap for various contributions.For being such a simple track, we found a lot of small things to improve on the previous BKT. Here's a rough breakdown:• 16 ms to a much better start slide and curve• ~35 ms on the first turn of laps 2 and 3, with a much tighter line plus an added slipchain (which was by far the hardest part of the run to TAS)• ~50 ms to a new shroom line, which sets us up better for the island and can afford much better boost distribution• ~85 ms to the islands; we accelerate a frame earlier from the rehop, avoid all bounces, and take a better line overall• ~12 ms to slightly improved trick and rehop• ~95 ms to a new last turn strat; we catch air and use mini-turbo deceleration to cut over a lot of off-road, while doing a slipdrift for realignmentEven though the track may look flat at first, the terrain is all bordered by shallow offroad slopes. With precise positioning, we can use these slopes to get air, allowing us to cross the water in many places without breaking the wheelie or slowing down excessively. The range of "good" positions is often very small though, much shorter than the length of movement in one frame; this behavior is we refer to as QM.Bad QM on this track loses a lot of time, so we're often forced to give up a small amount of timesave to fulfill conditions for good QM. The two major roadblocks are the first water puddle, which only works out in ~1 ms ranges out of 16.68 ms in a frame, and the islands section (particularly the slipdrift) which involves several variables to make it through without bounces and speed loss. On lap 3, after the islands there are no notable QM locks, so we went all out with optimization.
[MKW TAS] Luigi Circuit (No Glitch, Flap) - 21.325 Corvid 2024-03-28 | BKT cut: 31 ms (by ego, sorry)𝅘𝅥𝅮 Disturbed - Sacred LieThank you @MarthTAS for the new shroom strat, and @EjayB for the endless support and finding a trick improvement.This was my first go at making and polishing a TAS. Far from the most exciting run, but I think I did a good job optimizing, so I'm quite happy with the finished product. The new shroom saved ~24 ms overall, better boost distribution at the trick saved ~5 ms more, and the rest I squeezed out at the last shroom.
ingenuity (ctr nitro-fueled, time trials compilation) Corvid 2023-11-23 | some notes:spyro -- edge grinds at spiral don't save any time. the first clip at the end is the fastest spiral i could find, tied with martin's sl. the second clip is an accident but should be a timesavepark nmg -- the lap wr is particularly weak; first pols can be 6 frames fastertubes nmg -- 1.11.3 course and 22.3 (maybe .2?) lap doable, i did not grind this track for long. far left landing seems best at cut. 2nd hop landing is crucial, aim for end of downslope. don't stay tight after line to avoid speed loss, there is a small bump in the offroad. more hops at uphill is better, but awkwardgorge ur -- u-turning into the spike is great for more height, making the cut much easier (went from 0% hitrate to like 20% very quickly), but slightly slower. someone probably had this idea alreadyboogie -- hangtime after pad is very consistent for me. (the 19 f hop on l2 is rare.) requires a different approach for the turn, but i think it's a net gain even at wr level0:00 various laps4:53 spyro circuit8:17 coco park nmg10:08 roo's tubes nmg11:59 meteor gorge nmg+ur14:30 jungle boogie16:17 papu's pyramid nmg+ur20:28 tiger temple nmg+ur22:59 tiny arena
