SDA logo

Released in the United States on 24 November 1992 ("Sonic Twosday") and in Japan the next day, Sonic the Hedgehog 2 helped the Genesis/Mega Drive devour more of Nintendo's console market share. This time around, Sonic was joined by Tails, who only made himself useful as the second player character in the new 2 Player Versus mode.

Sonic2   Sonic2

Return to the Game List, the FAQ, or the Home Page.

Single-segment: 0:15:18 by Mike 'mike89' McKenzie

Get Flash to see this player.

Author's comments:

Thanks go out to:
The SDA Staff (there are officially too many of you now) - for keeping SDA awesome!
Nate - for sharing my love of Sonic, and developing tools to keep making the tech side of speedrunning easier

The Sonic Center - - Still the most comprehensive resource on competitive Sonic gameplay anywhere.
Suzaku, and especially Stanski - for the previous runs of this game, taking the game further than I ever envisioned
Aglar and marzojr - the game's TASing crew. Marzojr in particular is an encyclopaedia on the Sonic engine, his assistance was invaluable on certain tricks
Everyone else who's speedrun this game in the last year. TimpZ, TeeNTee, HDL, AlecK47, Tiberius, Naegleria, the list goes on...
Everyone who loves watching Sonic games get destroyed - this is why I do these, after all!

If you want to learn to run this game, start here:

Speed run history:
20/2/06 - 24:23 - Suzaku
19/11/06 - 20:02 - stanski
25/2/07 - 18:54 - stanski
19/7/07 - 18:12 - stanski
28/3/12 - 16:13 - mike89
4/12/13 - 15:18 - mike89

So, after the last run I didn't see myself coming back to this in a hurry, and, well, I guess it wasn't a hurry, but something bothered me about the old run. It just wasn't quite as good as I wanted, and in the interim a legion of runners had come storming after the old mark of 16:13, led by TimpZ.

In June I did a bunch of attempts, and improved that run by... four seconds. Not nearly enough, especially considering I'd thrown a new 12 second timesaver into the route. Not too long after, TeeNTee posted a video of a 16:03 on Youtube, which spurred me back into playing - as well as showing me a bunch of new strategies. From here, one of us breaking 16 minutes was inevitable, it was just a case of who cracked first. And if it was going to be me, I wouldn't be able to do it on the strength of my own strategies alone.

Finally, when presented with my uni holidays in November, I was able to have a strong crack at sub 16 again. Amazingly, no one had gotten it in the previous five months, and any run that was on pace found a way to die in one of the clutch levels at the end. One day in November I had a run tracking for 15:43 that died on hit 8 of the Death Egg fight, which stung quite a bit... for about an hour, after which I completed a 15:48. I thought I could take a break for a while after that... but I woke up the very next day to find out that TimpZ had choked away a 15:44 in Wing Fortress. So back to the grind it was.

In one night, I got 15:47 and 15:40 on stream, then a couple of days later, got this record off stream. I think it was best for me that I did, as I really noticed my concentration fading while streaming. As it turns out, this was by far the most consistent run I've ever done, and probably the run I can be most proud of out of any I've done in my 7+ years in speedrunning. There were very few mistakes, and any mistakes I did make I recovered from as best I could.

Oh, and guess what? I'm still playing on a black and white TV!

On to the run, then:

Emerald Hill 1 - 0:20 [±0:00]
Even the early game has changed a little bit. It's not easy, but you can squeeze one extra second out of all of the first four acts. I failed to get it in either act of Emerald Hill, but got them both in Chemical Plant. You can see some slightly more aggressive play compared to the 16:13, but most of it just comes down to whether you get the best possible spindash at the start. Here, I didn't.

Emerald Hill 2 - 0:38 [±0:00]
I actually started this stage about a half second late and I can't remember why! It's possible to get 0:37 here but I've already screwed myself out of it - the moving platform up ahead will be in the wrong spot and will block my jump, so instead I just slow down at the gap and save 38.

Chemical Plant 1 - 0:22 [-0:01]
Everything up until the block at 0:11 moving out of the way is completely irrelevant to the final time here. The keys to shaving the extra second from 23 to 22 here are the jump off the sinewave at 0:14 and the final spindash: you want to stop here because it triggers the blue snake up ahead but its also flat enough for you to charge a spindash. Then you have to time its release so you're not too fast and hit the snake on the way down!

Chemical Plant 2 - 0:39 [-0:04]
Pretty much exactly the same except for a jump off a ramp at 0:20 which nets you enough extra speed to shave the extra second. The large ramp jump at 0:18 is also more consistent now - TimpZ devised a method where you deliberately make the speed cap kick in just after making the jump, so you get all the vertical momentum but stop the horizontal, greatly widening the window to make the jump.

Aquatic Ruin 1 - 0:22 [±0:00]
Surprised to get 22 here, considering I didn't get a good blind jump at the speed shoes, but no complaints.

Aquatic Ruin 2 - 0:44 [-0:04]
Despite getting that tricky underwater jump (you'll notice it doesn't hit the pillar, but the ceiling just behind it - proving this jump is not at all frame perfect), this is probably the worst stage in the run, just for the boss. I had a method of timing the last hit that gave me relatively good odds on getting the boss glitch here - I oscillated between 1/4 and 1/2 success depending on whether I was having a good day - but on this occasion I somehow pressed right and kicked in the speed cap. In retrospect, it was very lucky the first arrow was shot at that height, as any other height would have cost me more time getting that last hit. This really should have been 42 at least, maybe even 39 if the boss glitch had worked.

Casino Night 1 - 0:28 [±0:00]
Really strange level, this. At 0:12 I missed the flipper shot causing me to land on the bumper, costing me quite a bit of time. I actually didn't think I was any chance of keeping the speed shoes all the way and was preparing to reset... but at 0:16 I managed to pull off what I believe is a frame-perfect jump similar to the yellow spring jump in EH1: propelling me at just the right trajectory to miss the ceiling and clear the side wall, making back the time I lost, or at least enough of it to make the last jump in the stage comfortably.
On the flipper at 0:23, I need to release right for a fraction of a second, this lines everything up so that (hopefully!) you fall through the flipper at the bottom. Missing this costs two seconds.

Casino Night 2 - 0:58 [-0:02]
TimpZ finally managed to convince me that the upper route is significantly faster here - there's probably about one second in it. Of course, when attempting to hit the top path I failed anyway, so whatever. 58 here is still a very strong time.

Hill Top 1 - 0:40 [-0:02]
It turns out just playing really damn well is enough to get the 0:40, you don't need luck at the seesaws. The most important part of getting this right is the ramp right after the S-tube: you want to hold left at just the right point so that you can hit the gap, but also clear the ball on the seesaw as you move back to the right. If you can do that, you're in good shape. But just to be sure, as I go through the underground section I look for the third door to close behind me just as the timer ticks to 0:30. If that happens, then I simply do this, and the rising platform here is in just the right position to ensure I dodge the lava.

Hill Top 2 - 0:57 [-0:05]
I saw TeeNTee go for the early cycle on this stage in his run. He failed it, but I wanted to see just how consistent I could get it. As it turns out, it's not too bad - and of course, there's no loss in attempting it if the worst result is just getting the normal cycle anyway!
If you pass into the underground section before 0:17, then the early cycle is yours. The platform you see moving upwards at 0:19 is the determinant of this: if you miss that you have to wait for it to come back down. You can actually catch it on the way back down and get a "half-early" cycle if you miss here.

Mystic Cave 1 - 0:33 [-0:02]
Lots of little new stuff that all adds up here. First I spindash under the pillar at the start, which I was always too chicken to do before. Second, taking the red spring at 0:20 saves a fraction of time (that one courtesy of TeeNTee), and the final spindash can be coupled with a jump to make the end consistent. I'd somehow completely failed to notice the ceiling here in all the years I ran this game. >_>

Mystic Cave 2 - 0:42 [-0:03]
Very simple stage, the only thing really worth noting here is the boss. My attack position is specifically chosen since very few spike patterns interrupt it. It didn't happen here, but if you see the first two spikes fall on the right side of the screen, then the left, you should move a little bit to the left because there's a spike about to fall on top of you.

Oil Ocean 1 - 0:34 [-0:02]
Pretty much the same as before, just cleaner. It's possible to save a second (or even two) by being more aggressive at some of the points where I stop in the oil, but it's very risky and with a run on this good pace, I decided to hang back a little here. Sometimes I still get 34 even when I throw those in, too, so it feels like a waste to attempt.

Oil Ocean 2 - 0:42 [-0:11]
So this is one of the bigger changes from the old run, favouring the level wrap here. This seemed to be straightforward for everyone but me; I got a very slow zip speed inside the wall making the zip impossible and I wasn't able to figure out why. As it turns out, the solution is to do something incredibly non-obvious: press nothing. As long as you're facing right, holding nothing actually pushes the spring in! So TeeNTee found that, but then there was still the matter of getting the zip correctly. It's not too hard even the old way (you need to hold right, then release it within four frames of exiting the wall) but AlecK47 discovered by accident that you don't have to let go of right at all with this method. So in the space of a day, this trick turned from the run-ender to 100% consistency!

Metropolis 1 - 0:58 [-0:09]
Don't be fooled by that time save, this level is still bad. Quite possibly worse than AR2 even...
This whole level is about making Tails do exactly what you want him to do. If you lose him at any point in the first three seconds, he'll come back as you spring up to the crab enemy... and the brown floor won't close beneath you, which costs a load of time. So it pays to take this a little bit careful. That made the cycle of moving platforms at about 0:14 much tighter... until I figured out another small timesaver. The first switch in this stage has always bugged me: the preferred way of handling it has always been to run into it and then jump on it. By jumping off the pistons and kicking in the speed cap, I was able to hit it without slowing down, which makes that cycle much more straightforward.
Now as we go up the yellow bumpers at 0:19, I see Tails - this means the best spot to make him land is at the bottom of this next part, on the right side of the mesh cylinder.
The next nut/bolt is why you need Tails to be back now - Sonic and Tails' speed stack on these. The first one went as expected, the second one... not so much. Apparently Tails really wanted to do a full jump, even when I didn't. That cost me quite a bit of time later, after the screen wrap. If the clock reads 0:47 as the gap to spindash through opens up, that means the third arrow platform will have the arrow facing up, which is disastrous. So here I have to hold my spindash back a little to clear it with the jump. Really this stage should have been around 0:54, but Tails making one bad jump set off an entire chain reaction costing me a few seconds...

Metropolis 2 - 0:16 [±0:00]
Yeah I paused, so what :p

Metropolis 3 - 0:45 [-0:02]
I paused here too. In entirely the wrong spot, both times - and still saved it!
The first pause should be a couple of frames later, as the platform is just starting to sink. Here I had to unpause and then start holding right a couple of frames later, then pause again. Again, I paused in the wrong spot - the right spot here is either right at the top of the ramp or right at the bottom - so I had to hold right while unpausing and let it go 2-3 frames later. Miraculously, both of these saves worked.
Tails getting hit 7 on the boss was a complete accident, but it may have actually saved a fraction of a second. We'll never know for sure...

Sky Chase - 2:06 [±0:00]
TeeNTee found a way to get 2:05 here, but it's not exactly consistent, and if you miss you actually get 2:07. So forget that.

Wing Fortress - 1:44 [-0:08]
The start is slightly quicker, the platforms played really nice, but the most important part of this stage is the end. Spindashing on the second last zipline just as it goes through the first ring sets up what we call the slope glitch, which completely breaks collision detection in hilarious ways. In this instance, I use the ramp near the end of the stage to push me down into a wall. That's been known for a while, but the real genius was contributed by TimpZ, who found that if you simply inch towards the boss room rather than going there in one fell swoop, you should be able to spindash to kill the boss -every time-. (Context: when slope glitch is active, the game thinks you're standing on an object with 0 width, so you're always at the edge of it - and when you're on the edge of an object you can't spindash. That, however, can be overwritten by any other object - including the one that the boss is! For some reason it only works if you inch into the room though, we don't really know why.)
Importantly, the boss always moves into the same position in the room, so you have to make sure you move into the right spot in the ceiling - the first vertical pillar in the background is the cue I use. You can also stop the very next tile to the left of this and it will still work.
Since slope glitch is still active, the cutscene is broken - the game's trying really hard to make Sonic look up, but you have to be standing still to look up, and slope glitch doesn't permit that in normal circumstances. So this happens.

Death Egg - 0:50 [±0:00]
It's a wonder my controller even worked during this stage. I couldn't keep my hands still!
In the end this went exactly the same as the thousand other times I've beaten this stage, so no harm done, and the game is over!

Total time: 15:18 [-0:55]

With a time like that, it's amazing to think there's still more to be had... but there is. Good luck to the crazy soul who attempts it, though.

Enjoy the run!

Single-segment with Knuckles: 0:15:58 by Mike 'mike89' McKenzie

Get Flash to see this player.

Author's comments:

Thanks go out to:
The SDA Staff (there are officially too many of you now) - for keeping SDA awesome!
Nate - for sharing my love of Sonic, developing tools to keep making the tech side of speedrunning easier, and providing the capture card used for this run

The Sonic Center - - Still the most comprehensive resource on competitive Sonic gameplay anywhere.
stanski - for many, many strategies. He claims he sucks with Knuckles, but there are a bunch of times he has that I still can't touch.
werster - even more help with strategies; a lot of his stuff is not exactly SS-safe but its great to see just what is possible in the stages
Fuzzerd, Aglar, Upthorn, Sonikkustar - the game's TASing crew. Shoutouts to Aglar in particular for explaining the finer workings behind the more difficult tricks in this run
DMTM - for discovering the Casino Night 2 level wrap (this is actually an anti-shoutout because this trick sucks)
And last but not least, MrSparkle, for discovering the clip through the mesh in Metropolis 1, which turns a potential stumbling block into a relatively easy level.

After I finished the Sonic run I decided I had had enough of speedrunning for at least a couple of months. I was going to sit back and just play games casually for a while. The thing is, when the speedrunning bug bites you, that's not how it works. My desire to do this run was rekindled in a couple of different ways: firstly, I have to thank a couple of guys over at TASvideos, Felipe and skychase, who took on a project I had considered a while ago, completing this game with Knuckles while not collecting a single ring, and succeeded at it! I couldn't help trying out some of the tricks, and especially stuff like Casino Night 2 helped me become familiar with the more difficult stuff in the run.

Around this time as well, I was streaming this quite a bit, and consistently getting high viewer counts (up into triple figures some nights!) so I figured it was something that people wanted to watch. The more I thought about it, too, the more I reliased that this run would be significantly different from a Sonic run, even though the final times would be very similar. And with that I started working on the run for real.

One general thing about this run (and probably the one thing I'd change about this game if possible): unlike Sonic 3, when Knuckles drops from a glide in Sonic 2, even if you're holding down you can't charge a spindash until the "I dropped from a glide" animation ends, which takes 15 frames, or 1/4 of a second. So if some of my spindashes look slow, that's a possible reason for it.

Some terminology that I'm likely to use throughout my commentary:
* Frame - 1/60th of a second, the smallest unit of game time.
* Frame window - the amount of these units you have to execute a particular action. A "frame-perfect" action has a window of just one frame, for instance.
* x-tap spindash - Each additional button press in a spindash gets you more speed, up to a maximum of six. Sometimes having a specific amount of taps gives you the right speed to get through a certain area quickly or easily.
* Universal timer - There's two different timing properties of moving objects in this game. The universal timer is where an object starts moving the instant you start the level, and will therefore be in a certain position at a certain time, every time.
* Camera timer - The other kind is this one, which only starts moving once it's brought onto the screen (or near the screen - when the object is first rendered, in any case). Knowing which objects have which timing properties is important to good times when you have to deal with them.
* Ramp jump - Knuckles's jump speed and height are dependent on both your speed going into the jump and the slope of the ground you jump off. Running up a slope causes your jump to become much higher, while running down a slope will make you jump quite flat to the ground, gaining speed. This applies even if the slope is very small.
* Loop jump - Similar concept, but inside a loop. If you're moving through the loop from left to right, a loop jump is from the top-left to the bottom-left. It's kinda finicky because if you land at different angles you get different speeds, but in the best case it is faster, and it's also a very convenient way to switch from rolling to running.
* Superglide - Simply jump on an object (an enemy or an item box) and glide just before you hit it. This causes you to translate your falling speed upwards, and since you're gliding you have a much lower rate of acceleration towards the ground. In layman's terms, you end up a lot higher than a standard jump.
* Path switcher - This is an invisible item in the game that allows the levels to work as structured. There are three different levels of background/foreground and Knuckles is always in one of them at any one time, and these items seamlessly switch Knuckles from background to foreground and vice versa. This is especially necessary for things like loops to work, because if you didn't have them they would just be solid all the time and you wouldn't be able to enter it. Knowing where these are and how they work is key to some tricks in this game.
* Zip - A zip occurs when Knuckles is stuck in a wall somehow and presses left or right. Since collision detection is designed to stop you from going into a wall in the first place, pushing left against a wall makes the game push you back to the right so you aren't embedded in it. As a result, when you're already inside a wall holding left or right causes you to travel in the opposite direction much faster than you can attain by normal play.
* Level wrap - This is what happens when a zip causes you to bypass the left edge of the screen. The game treats your X-coordinate in memory as a number less than 0 - but position coordinates are unsigned, so instead of a negative number it's a VERY large positive one - which correlates to the end of the stage. In this run Casino Night 2 and Metropolis 2 are skipped this way.
* Screen wrap - Similar concept, but using Y-coordinates instead. If you can move off the top of the screen, the game thinks you're all the way at the bottom. This only works in certain levels which wrap around themselves infinitely - which in Sonic 2, is the three Metropolis stages only.

On to the run, then!

Emerald Hill 1 [0:20]
[0:06] About the only thing that differs from a Sonic run in this stage is that this jump over the spring is much tighter due to Knuckles's lower jump. Other than that it's just as straightforward.

Emerald Hill 2 [0:38]
[0:14] This is a slightly trickier level for Knuckles because of this ramp jump. What I need to do is bounce off the invincibility and carry that bounce across to the bee enemy up ahead. The problem is that Knuckles can't regulate the height of his bounce - once you let go of your jump, pressing any of them again will cause you to glide instead. Here I pulled the jump off with the correct height, but it's very possible to go flying way too high off the ramp, which means you spend too long waiting before hitting the invincibility and costs you a second.

Chemical Plant 1 [0:23]
[0:18] This is the one place where the speed cap is actually useful: Sonic just holds right and jumps straight into this gap, whereas Knuckles has to slow down and line it up.

Chemical Plant 2 [0:40]
[0:05] Yes, it's possible to catch an early cycle here by just jumping off this ramp and hoping for the best. However, that's exactly the problem, you're just hoping for the best, you don't really have any control. If you jump straight off the ramp there are no less than five different possible outcomes:
a) Land on the platform and continue running
b) Land on the platform and stop dead
c) Land slightly embedded in the platform
d) Land completely embedded in the platform
e) Fall through the ramp and die
Of these, a) and c) are the results that can get you to the early cycle in time, but they are both quite rare. The most common result is b), followed by e) which is obviously the worst. If you get d) you can zip to the right but the door will kill you, so you can't make the early cycle.
After e) started happening to me a lot more than usual I decided to just take the loss and do it this way. It's equivalent to getting b) or d) from the list above and gets more attempts through to Casino Night 2, which is the most difficult trick in this run.
[0:24] If you clear this platform you can save a second, it's a tricky jump with Knuckles just because of his lower jump but you can make it kinda consistently.

Aquatic Ruin 1 [0:23]
[0:04] Screwed this up a little bit, I got too close to the enemy and my right hand panicked a little and went into spindash mode early
[0:19] And then completely missed this jump up to the middle of the loop, which cost about two seconds. No big deal though.

Aquatic Ruin 2 [0:41]
[0:11] Superglide off this box -> make stage entirely trivial. It needs to be a pretty specific height though: too high and you bump into a pillar near the end, and too low and you catch the lower of the rotating platforms, and gliding gets faster the longer it goes, so there's no way to catch that up. High is fine since you can just stall a bit to lower your height (which probably costs a second), but if you're too low you lose probably four seconds.

Casino Night 1 [0:29]
[0:14] This jump is actually much easier as Knuckles because of the smaller jump. A full size jump from the shield box here goes straight through without hitting any of the bumpers (or hitting the last one in such a way that it doesn't matter).
[0:21] Knuckles can't make the jump across to the top in one go so it's slower than just taking this route.

Casino Night 2 [0:35]
Here we go. This is THE level, the reason why Knuckles is faster than Sonic, the reason this run took as long as it did, the reason Aglar gets such a shoutout in my thanks up above.
[0:04] OK, from the top. The first part of this trick is to do a full size jump off the top of this platform and do a frame-perfect glide to clip behind the speed shoe box. Too early and you have to climb back up again (this takes about five seconds, I normally gave this two attempts), too late and you break the box and the run's over.
[0:05] Now, before I go through the rest of it, if you're not familiar with TASvideos, allow me to blow your mind by introducing the concept of a sub-pixel. Yes, there are smaller units of measurement in games than pixels! In this game each pixel is divided into 256 sub-pixels, and having these in the right range in both the x- and y- planes is essential to this glitch working. I'm not even kidding. The good news is, you can manipulate your x-subpixel into the right position. The bad news, you can't do the same for the y-subpixel. So, once behind the box, I have to tap right for 2 or 3 frames, this causes the x-subpixel to be in the right place. Now I spindash and get pushed into the floor, and do two more quick spindashes. This is where I get to find out whether my subpixels are good.
[0:07] Obviously this is the intended result, but if my y-subpixels are no good I will be pushed down into a box in the wall that I can't escape from, and if my x-subpixels are no good I will be ejected much more slowly than this (not nearly fast enough to go through). About 3/8 of the potential y-subpixel values will not work, and it can't be manipulated, so I only have a 5/8 chance of pulling this trick off even with the correct input. It's the closest thing to a legitimately "random" element in any of the classic Sonic games.
But it's still not over! I also have to let go of right at the right time. There's two windows for this: one is the first two frames after exiting the wall, and the second is three frames just as you go off screen. I completely whiffed the first one but luckily paused in the right area for the second.

Hill Top 1 [0:40]
[0:14] There's a tiny part of this ramp that's much steeper than the ramp around it. Hit it and you get this jump all the way to the top of this wall instead of halfway up.
[0:23] Same as Sonic's zip, though I got really lucky here and ended up going slightly further than usual, allowing me to catch the platform instead of having to bounce back up off the spring.

Hill Top 2 [1:01]
[0:04] Probably the biggest mistake in this run, my spindash wasn't quite strong enough and I failed to clear the loop. If you're wondering why I didn't glide to try and climb up, you can't grab that part of a loop. Cost close to 4 seconds.
[0:20] This superglide is very handy because it means that you're not reliant on the cycle of moving platforms up ahead at all - the path across the top doesn't have any.
[0:32] This jump is way easier as Knuckles because of - you guessed it - that short jump again!

Mystic Cave 1 [0:36]
This is my favourite route in this run. I was streaming one night and werster pointed out a bunch of improvements and simplifications, so I ran the level over and over on stream trying to nail it down. It's one of the more technical levels in the run, and I'm really glad with how it turned out!
[0:15] This is the first change - Knuckles takes the top route here. This would probably be faster with Sonic too if it weren't for the damn spikes at the top of the ramp. From here you can spindash and jump, land on the box and jump again, to skip that pesky lower route.
[0:23] Now the second part. We used to superglide off the invincibility box on the lower level, but I realised that you could just use this enemy instead. It's a ton easier and roughly as fast - the only danger is if you glide to exactly the wrong height, you hit the bottom corner of the wall and drop off.
[0:30] Finally, instead of making the last bridge fall you can just climb up to the vine, saving about half a second.

Mystic Cave 2 [0:44]
[0:12] Knuckles can't jump up to this platform, so instead we get to see the only application in Sonic 2 of a damage boost!
[0:33] There's five different patterns I've observed for the way these spikes fall - as soon as I saw the one fall through the middle I knew that standing on the right was completely safe.

Oil Ocean 1 [0:36]
[0:08] The only mistake in the stage was here, didn't spindash for some reason. Apart from that this stage was pretty much flawless.
[0:26] From here, five full jumps gets you in position to quickly climb up this wall and spindash away before the fan kicks in. If the fan stops you you lose a good four seconds waiting for it to stop.

Oil Ocean 2 [0:55]
[0:14] Changed my strategy on this slightly, instead of trying to line up a 4 or 5-tap spindash I just do a full 6 and then slow down just before hitting the ramp. It has the same effect.
[0:24] Another slight change, Knuckles can't hit the spring in the background, since it's too high up, so he has to take a hit on the spikes and use them to get up.

Metropolis 1 [0:49]
[0:28] Shoutouts to MrSparkle! He found this trick to clip through the mesh cylinders. You can also do it on the first cylinder but it only saves about three seconds if you get it first go, and it's frame-perfect.
[0:40] If I had got here half a second earlier the last arrow would have been pointing up and that would have been bad news. But as soon as the second one retracted I knew it was safe to go.

Metropolis 2 [0:16]
This is very much the same as with Sonic, except as Knuckles you glide into the zip point, which is easier... unless you're at the end of a speedrun and like choking like I do. >:/ Again, once I activate the zip I pause to line it up. This one has a four-frame window to release right, so it's significantly easier.

Metropolis 3 [1:20]
This level was PERFECT for my route. Stanski has a time one second faster, which is doing everything here plus spindashing through the mantis at the start of the stage instead of jumping over it, which is about the most SS-unsafe strategy ever.
[0:11] Knuckles can't quite do this glitch in the same way Sonic does. Luckily, he has a much easier method.
[0:22] Holding one of the jump buttons causes you to go much higher up these. You can use this knowledge to take this completely safe path over the mantis at the top.
[0:40] This is much riskier than I used to do here but I have a consistent method to do it now. Once I jump off the screen I'm watching a set of flashing lights in the upper left of the screen, about in line with the numbers in the HUD. Once all three of them move off the screen I jump and the bolt pushes me into the floor. If you're too late on this jump, you die, and if you're too early you just appear on top of the bolt, costing 5 or so seconds while you set it up again.
[1:13] I changed the last two hits slightly, attacking from the left side instead of the right. The reason is that by the time Robotnik hits the bottom of the screen, the last sphere is on the right if you've done the rest of the fight perfectly.

Sky Chase [2:05]
As with Sonic, this stage is simply free points. I screwed up the combo towards the end but it doesn't affect my time so I don't care.
For some reason, this stage ends a few frames earlier as Knuckles, so you can scrape a 2:05 in. Why? Who knows...

Wing Fortress [1:56]
[0:09] I missed this jump over the chicken here, so I have to do the next area slightly differently. This means that the last platform I need to jump off is in a slightly different position, but I've practiced for it so I know what's going to happen.
[0:19] This is kind of scary because I don't know exactly where the conveyor platforms up ahead are going to be, and so I'm really nervous about starting the glide in case I tag the edge of one of them, break the glide and can't recover. I ended up getting the glide off, but I was quite low to the ground and, now I think about it, was very lucky that one of the platforms up ahead was up.
[0:27] I'm not entirely sure if the old method I was using on this stage is even possible with Knuckles (spindashing up the ramp and jumping off at the very edge, as in the Sonic run), but werster pointed out to me that there is actually a hook over here! Interestingly, this method is just as fast as pulling off the tricky jump.
[0:34] And somehow completely missed this jump off the zipline. Lost two seconds to that.
[1:05] This trick is actually just as easy as it looks. Hold down, one-tap spindash, jump, pass through laser. Knuckles actually saves time with it because, unlike Sonic, he can glide to halt his momentum and fall down the gap.

Death Egg [0:51]
[0:13] Contrary to what I thought earlier, Knuckles actually -can- trigger the infinite Silver Sonic glitch which locks the camera in the Silver Sonic arena until time over. He also has a second possible glitch, where after the 8th hit he doesn't lose his hitbox and can still kill you (or you can kill him again for 100 points), which I thought replaced the infinite Silver Sonic glitch, but I discovered while dicking around before an attempt session that that wasn't the case. So I just play this fight really safe and make sure that I don't kill him anywhere near the glitch frame.
[0:26] Every single one of these hits is a nerve-shredder at this point (even hit 2, which should be completely free - I've had Knuckles simultaneously jump into the boss and the flame and kill me before). Prior to this run I lost two (slightly worse) runs on hit 3 and hit 6 respectively. The relief once he flew up after hit 9 was immense, and my reaction after escaping was... slightly excited. :)

I have no doubt there is still improvement to be had in this run. My best attempt that got ruined late would have come out to 15:51, and with these strategies, Hill Top 2, Metropolis 1 and Wing Fortress have five seconds worth of improvement each on this run. If I hit the Chemical Plant 2 early cycle as well, that's another six seconds, putting the run into 15:3x territory. However, I wish whoever goes for that run the very best of luck. :)

Enjoy the run!


Get Flash to see this player.

Level Time Date Player(s)
Emerald Hill Zone, Act 1 0:00:19 2006-12-17 Joe 'stanski' Stanski
Emerald Hill Zone, Act 2 0:00:35 2015-07-24 'AlecK47'
Chemical Plant Zone, Act 1 0:00:20 2013-06-14 'Paraxade'
Chemical Plant Zone, Act 2 0:00:39 2007-06-21 Joe 'stanski' Stanski
Aquatic Ruin Zone, Act 1 0:00:21 2007-03-10 Joe 'stanski' Stanski
Aquatic Ruin Zone, Act 2 0:00:38 2013-06-19 Mike 'mike89' McKenzie
Casino Night Zone, Act 1 0:00:25 2015-07-23 'AlecK47'
Casino Night Zone, Act 2 0:00:57 2007-06-21 Joe 'stanski' Stanski
Hill Top Zone, Act 1 0:00:39 2007-01-10 Joe 'stanski' Stanski
Hill Top Zone, Act 2 0:00:56 2007-06-21 Joe 'stanski' Stanski
Mystic Cave Zone, Act 1 0:00:32 2007-05-04 Joe 'stanski' Stanski
Mystic Cave Zone, Act 2 0:00:42 2007-03-10 Joe 'stanski' Stanski
Oil Ocean Zone, Act 1 0:00:33 2007-01-04 Joe 'stanski' Stanski
Oil Ocean Zone, Act 2 0:00:40 2008-05-03 Joe 'stanski' Stanski
Metropolis Zone, Act 1 0:00:53 2007-06-21 Joe 'stanski' Stanski
Metropolis Zone, Act 2 0:00:14 2009-05-19 Joe 'stanski' Stanski
Metropolis Zone, Act 3 0:00:45 2012-03-27 Mike 'mike89' McKenzie
Sky Chase Zone ------ ------ ------
Wing Fortress Zone 0:01:38 2015-08-07 'AlecK47'
Death Egg Zone 0:00:42 2013-06-17 'Paraxade'

Joe Stanski's comments:

Emerald Hill Zone, Act 1 (0:00:19)

Emerald Hill 1: The difference between me getting :19 and :20 with this route is basically spindash speed. If you do the exact same thing as the TAS route (good luck with that), you might get a low :19, but I really doubt :18 is possible.

Chemical Plant Zone, Act 2 (0:00:39)

Chemical Plant 2: Just optimized the end a bit, very very close to the second border, I can't tell you how many 40.00001's I had.

Aquatic Ruin Zone, Act 1 (0:00:21)

Aquatic Ruin 1: If I spindash faster at the beginning, I could get :20. The rest of the route is optimal.

Casino Night Zone, Act 2 (0:00:57)

Casino Nights 2: I know for sure I can get :56, as I've done it with death abuse, but I seriously got sick of either having a slightly slow boss fight (here) or getting like 57.00001. This level is one of my favorites though, and one of the toughest to optimize, so its definitely a good watch.

Hill Top Zone, Act 1 (0:00:39)

Hill Top 1: This level I am very confident will not be easy to improve. If I glitch through the lava like I did in my 18:54 SS, and glitch underneath the earthquake (i.e. you can spindash-jump into the earthquake area and there is an area where the collision box is not there and you can pass under the ground for a short distance instead of having to break that rock), you might be able to save time, but I really doubt the rest of the level would go as good as mine did. The jump after I take damage and spindash has 1 frame accuracy, which is very fun to try to do correctly..

Hill Top Zone, Act 2 (0:00:56)

Hill Top 2: As you can see, I finally made it to the previous cycle of the rising blocks. I used to find it very hard to do that, but I get it somewhat consistently now. There really are no long stretches at the beginning to spindash, so it looks kind of slow, but it isn't.

Mystic Cave Zone, Act 1 (0:00:32)

Mystic Caves 1: I figured this out and there is a possibility for 10-12 time saving spindashes (most with jumps requiring a lot accuracy following them) in this level. Problem is, all of these spindashes save about a quarter to one half second. So, I got sick of missing a few, and picked a few that I wasn't screwing up as much. Skipping the invincibility and just damaging off the spikes would save another .75-1.00, but honestly I got sick of perfect runs being ruined right at the end, which is the spindash that saves the most time. :29 is probably the best with the routes I have in my head, but this level is a bitch and pretty good at :32.

Mystic Cave Zone, Act 2 (0:00:42)

Mystic Caves 2: I think this level is optimal for this route, the TAS route actually wastes more time then it saves for real players.

Oil Ocean Zone, Act 1 (0:00:33)

Oil Ocean 1: You can skip all 4 of those pillars by spindashing under the first 2, jumping, and then rolling under the next 2. I skip the first one, rolling under it instead. I don't think this wastes any time at all, and would be highly surprised if anyone ever beats this time. If someone could get the timing down on the jumping at the bottom of the platforms like the TAS does, you could really improve this, but I think that is basically TAS only ground. I'd consider this my hardest IL time to beat.

Oil Ocean Zone, Act 2 (0:00:40)

Oil Ocean 2: well, here it is, the reason i was so against OOB on SDA. I don't actually know if this level can be improved, but i'm pretty sure its close to optimal. It doesn't really matter, as the scrolling glitch is so hard to pull off in a run, and the boss battle after is also very hard, but i don't think anyone will ever beat this. Feel free to prove me wrong.

Metropolis Zone, Act 1 (0:00:53)

Metropolis 1: I'm glad Maur told me to improve this, as it is now one of my best IL's. Very smooth, and I kind of figured out how to manipulate tails somewhat consistently.

Metropolis Zone, Act 2 (0:00:14)

Metropolis 2: The zip at the start destroys a pretty cool level, so i'd recommend watching the previous IL if you want to see a more skillful route. But getting 14 took me a fair amount of time, because you need to start holding right at a very specific time. There are 2 frames of you can hit to get inside the wall, so yea the odds of hitting those correct frames aren't great. There is another method to do this glitch using the other platform but this one is easier.

AlecK47's comments:

Emerald Hill Zone, Act 2 (0:00:35)

The only real difference between this 35 second IL strat and the more Single Segment friendly 36 is the slope jump just after the speed shoes that skips two loops (all credit to TeeNTee314). There are two non-consecutive frames to get the slope jump, but only on the second will you have enough height to bounce on the monkey. Beyond that, most of my failed attempts were because I either hit the ceiling spikes heading into the bottom route and lost my momentum, or slowed down too much trying to avoid them. I very nearly slowed down too much in this attempt, but fortunately still got the 35 by a very slim margin.

Wing Fortress Zone (0:01:38)

For a while, this level had me calling the game "Lemming the Hedgehog 2" because of how many times the platforms were in the wrong position. Eventually I figured out that by making sure the platforms are on a fresh cycle (only jumping on the first platform after it appears, and waiting for it to leave and reappear if it's already there) and doing a three-tap spindash off the very same platform I spindash from in this IL, the timing of when Sonic dies would set up the cycle perfectly. After that discovery I got two 1:39's and finally this run in very quick succession. Although I very nearly lost this one, because for the last tap on the zip to trigger the boss I didn't fully depress the dpad. Fortunately I realized what happened, triggered the boss and finished the stage just in time.

Casino Night Zone, Act 1 (0:00:25)

This one was annoying.  Random as well, fittingly enough if one considers the level.  Basically you use the speedshoes to trick the game's physics so you run along the top of the screen, enter the wall, zip (or more likely get knocked out of the wall normally or softlock) and hope your subpixels are right.  Then once the end of the level is near (assuming subpixels didn't stop you halfway) press right, then left to trigger the signpost and allow the level-end sequence to play out, respectively.  Yeah - it's kinda messed up.

Three main things could be faster about this: jumping to the spring after the ride on the block, bouncing off the flippers and hitting a slightly earlier zip point (which doesn't have any good visual cue, unlike the one I used here).  All together those improvements could just barely get a 24, but I have no interest in grinding that out.

Paraxade's comments:

Chemical Plant Zone, Act 1 (0:00:20)

This was a very good run. The hardest part is catching the blocks, which requires a near-perfect beginning (though I messed up the slope jumps at the start a little bit and still made it). There's a zip that was found by Tee-N-Tee shortly after I did this run that can take it down to 0:15, but 0:20 is as good as it gets with this strat.

Death Egg Zone (0:00:42)

I think the difficulty of the single-cycle strat on this level is a bit overrated, it only took me a day or so from basically picking up the level with no experience to pulling off the 42 :P Emulator practice and advice from werster certainly helps though, thanks to him for that. If anyone's interested in knowing more about how to pull off the single-cycle strat, I wrote a guide over on TSC that explains it in more detail.

Mike McKenzie's comments:

Aquatic Ruin Zone, Act 2 (0:00:38)

I hit this in a single-segment attempt, but lost the recording of it. The next day, I set myself to spend a while recording attempts at this time, and got it in a single attempt.
This time is optimal for the route chosen. There's a known faster route that can yield 37 (or maybe even 36) by taking the springboard up to the top after the underwater jump into the pillar, but every time I try it I just get flung into sharks. So... good luck?

Metropolis Zone, Act 3 (0:00:45)

Metropolis 3: Could probably be a bit faster with a spindash at the start and less inching towards the zip point, but it was quick and the boss went well and I didn't expect 45 anyway.

Return to the Game List, the FAQ, or the Home Page.