Hi, this is my Shadows of the Empire speedrun. I think it's pretty fast.

I initially started speedrunning this game intending it to be a side-project, with no ambition of putting it on SDA. I destroyed my first goal though, and decided to lower my goal until I got a time I thought hard to beat. I didn't even consider myself capable of 46 until well into my attempts for 47, but Alaris suggested that if I were to get a run that good then I should try to update the run on SDA.

Now that I've achieved my goal in this game, my focus will be on growing the community. If you're interested in speedrunning Shadows of the Empire, read this guide and feel free to contact any of the players mentioned there if you have questions. We hold regular races on SRL that you are welcome to join, which are usually followed up by novelty races. We're also working on a YouTube channel for tutorials, game theorizing and community highlights, so check that out!

I play on the Japanese version, which is inherently slower for this category by about 5-10 seconds. I bought the version most convenient for me, thinking it wouldn't matter since I didn't intend to get an amazing time. By several strokes of luck, it turned out there are many subtle changes that make each version feel totally different to play, yet the advantages evened out enough to make it a matter of preference. I chose to continue using the Japanese version because that's what I'm most comfortable with.

46 minutes real time was my ultimate goal time. Though this run is improvable (as all runs are), I will never be able to beat this easily unless a huge break is found. If I were to get the best run I could reasonably hope for, it would be around 30 seconds faster than this, and I lost more than half of that to luck factors. I'm very happy with this run and I'm ready to move on.

This game has come a long way since I started running it, so I'll do my best to explain what's going on and what's changed. I'm very sorry about the text wall, but I wanted to include as much information as possible.

Battle of Hoth — 2:00

Acceptable, but not great. This level is extremely difficult to get a great time on thanks to the extensive efforts made by the community to optimise the route, the heavily RNG-reliant nature of the route, and of course the difficulty of executing the level precisely in the first place. On top of that, mistakes on this level can easily snowball into a huge time loss due to there being no way to quickly correct yourself (you can't stop moving). Because of all that, I decided to lower my standards a little on this level for the sake of getting runs going.

Though it may not look it, this level has gone through the most drastic route changes. The realization that stages won't complete until the death animations of all enemies are over has had the greatest impact. Probes die very quickly, while AT-STs and AT-ATs die slowly, meaning that Probes are left for last unless it would require a huge detour. This is most obvious in the strategy for stages 2 and 4. Finally, the way the AT-STs and Probes move has a large impact on how tight your line through the level can be.

This level has no known advantages on any version for Any%, and for 100% the advantage is insignificant (a less than ideal camera order when exiting tow cable mode on the Japanese version). It has been speculated that one version might lag less, though opinions are mixed on which version that actually is, and no in-depth test has been made.

Escape from Echo Base — 3:22

Very good, only 4 seconds slower than my "RTA strats" best despite getting an extra pack of seekers. This level has the first instance of abusing "cruise control", which is offsetting the center-point of the control stick to send larger values to the game in one direction (achieved by tilting the stick while pressing L+R+START or plugging it in), resulting in faster running speed. Aside from that, this is a very straightforward level, the main failure points being the risky jumps over gaps, made much riskier by the fact that this version has smaller jumps.

The main routing change on this level is fighting the AT-ST with only lasers, abusing its weak point and saving yourself the extra trip for more seekers. The exact seeker packs you pick up has gone through several iterations too, now ending the level with 3 packs (15 seekers) for use on Ord Mantell Junkyard and Gall Spaceport.

English versions are slightly advantaged on this level due to larger jumps allowing you to cut corners better. However, Japanese compensates a bit due to less lag and more laser damage at lower power. Overall, English is about 2 seconds faster.

The Asteroid Field — 3:25

A fairly average time, but acceptable. This level consists of killing 60 TIE Fighters and 20 TIE Bombers. Fighters will usually appear in packs of three or four and Bombers in packs of two or three. Missiles are in almost every situation the fastest way of killing packs. The problem is that the pack order and flight pattern is random, you don't always have missiles to use, and Bombers are much harder to hit with lasers. Like Hoth, I decided to settle for less on this level for the sake of continuing runs. An amazing time on this level is mostly about RNG, but you can still get around 3:30 with the worst RNG and good play, which I set as my limit. A combination of average play and average luck resulted in an average time barely below the cutoff threshold.

This level is probably the easiest to see how RNG plays its part, but the level is actually not as luck-based as it appears. The level is actually an auto-scroller, and TIEs spawn in certain spots along a predetermined path, forming a queue. Normally you would kill these packs as fast as possible and force the queue to catch up to you, but because you travel to the packs yourself, screwing up the packs at the start doesn't matter that much as long as you've caught up on the queue by the end. Also, though random, TIEs move in predetermined ways, and the randomness comes from which pattern it picks. This is most significant with the Bombers, because if you learn to recognize which pattern they're on, then lasering them is much easier (though still hard).

Another level with no known version biases, and one of the two levels with no known version differences at all.

Ord Mantell Junkyard — 10:23

Great time for the Japanese version; only 2 seconds slower than my best. Junkyard is essentially a 9:30 train ride followed by a short fight with IG-88. Nothing noteworthy ever happens on the train section because it's an autoscroller; it cannot be sped up at all aside from saving about 3-4 seconds in lag reduction, and a lot of work has been put into figuring out exactly where the best spots to look at what part are. The fight with IG-88 went well. I got the second-best movement pattern; he went to the ideal spot but moved a bit faster than ideal, causing me to miss a couple of seeker shots. Still fast though, and had a few seekers left over for Gall Spaceport too.

Though fairly straightforward on English, IG-88 is by far the most complex single thing in the game on Japanese. He moves faster, makes his decisions at shorter intervals, and has a larger variety of locations he will choose to go at any point. However, the time used to get the extra seekers on Escape from Echo Base is immediately saved in full on the fight with IG-88. Aside from straight up speeding the fight up in the best case, it also made the fight much safer because having higher damage output gives him less chance to escape.

English is slightly advantaged on this level due to the fact that seekers do more damage and IG-88 plays nicer, making the boss fight quicker. Japanese is advantaged in the train section, as you can lag reduce one of the laggiest sections much more as a result of taking less barrier damage. Overall English is about 4 seconds faster, though on average the difference will be more due to IG-88.

Gall Spaceport — 7:31

Decent time for RTA. This is the level that defines speedrunning Shadows of the Empire. Excluding Hoth, the vast majority of my resets were because of this level or as a direct result of it. It also fully showcases the game's largest glitch: jetpack clipping. Generally speaking the level as a whole went really well. I lost almost all the time I did here due to luck with the boss clip and the Boba fight itself. Fortunately, the extra leftover seekers from Junkyard helped keep him under control. Slave I went very well, as I was able to save a lot of pulse. I actually use three different controllers in this level, because touchier sticks make sideways movements (such as swaying with the jetpack and strafing) less of a pain, but at the cost of general movement accuracy. I switch at strategic (hopefully invisible) points to avoid losing time.

Gall Spaceport now uses a new skip to get to Boba that saves roughly 1:30 and was previously thought to be TAS-only. The trick to do this is notoriously temperamental, and with current knowledge no amount of practice will make you consistent, despite the many improvements to the setup. A new method for skipping the AT-ST also saves a fair amount of time here. Boba, like IG-88, makes his decisions faster on Japanese and he also dies slower due to seeker damage. Though a perfect fight should be almost identical to English, this is pretty rare, and usually you just end up fighting him in the air. Luckily my aerial battle was far from the worst it could be.

There are a lot of differences to this level, but it more or less evens out in terms of time impact. On the Japanese version, most weapons do less damage, but enemies do less damage too, making it easier to run through the level without problems. The fight with Slave I is sped up due higher Japanese pulse damage. On the other hand, English has an easier and slightly faster Boba fight due to higher seeker damage, and can use freefalling to save time more effectively in the jetpack sections. All of this is a matter of a second or two.

Mos Eisley and Beggar's Canyon — 2:37

Excellent time, especially for the Japanese version. This is the level I'm probably most happy with in the run. It has a lot of subtleties if you want to go fast, and aside from a few unavoidable bonks it all went smoothly. The idea to this level is that you're supposed to beat all the bikers in order to finish. If you don't, then they all shoot off into the distance and you can't catch them. Playing on Easy allows me to do the biker glitch, where you only kill the first guy and take a particular path through the level. This causes them all to slow down and let you pass after a certain point, saving a non-trivial amount of time.

It's irrelevant for Any%, but the other major difference on this level is that half of the challenge points are in different locations. Overall they are in much better spots for speed, except one of them is in the path you need to avoid for the biker glitch. Because the glitch is difficulty-exclusive, this placement actually makes Japanese 100% slower on Easy difficulty, but faster on higher difficulties.

In theory, there should be no difference between the best Japanese and English time on this level. However, Japanese is made much more difficult by the fact that it's both possible and easy to die, comparable to Jedi difficulty on English without the darkness. Even slight bumps or touching the ground beneath you can cause you to die. On the plus side, you get real good at this level real quick as a result.

Imperial Freighter Suprosa — 2:49

Nothing special, but an acceptable time. Generally speaking, this level went reasonably well. Clipping the red door went poorly, but like the boss clip on Gall Spaceport it's not consistent with current knowledge, so I'm willing to accept couple of failures. Jumping over the barriers the "wrong" way in the conveyor room went pretty smoothly, and I made it to the storage area with above average health. Loader Droid wasn't the best it could be, but it was still good enough to spare enough pulse for Sewers, which is uncommon.

Using pulse exclusively is a fairly recent development, even for the Japanese version (I used to use pulse on Loader, and flame in the level itself). It's a lot faster, but in the context of a full run makes ammo management much more difficult. Even so, the time save is enough that the better English players are skipping flame now, despite the small pulse disadvantage.

Suprosa is advantaged on Japanese primarily because pulse does more damage. Taking less damage in general also has the largest effect on this level, meaning you can run past enemies freely and will almost never need to make a strategic checkpoint death. Some of the moving parts on this level also move slower, giving you better movement cycles and posing less risk. English does have a "better" conveyor skip, but it's unknown if it saves time. Overall Japanese saves about 1-5 seconds depending on an English user's choice to get flamethrower.

Sewers of Imperial City — 3:18

Decent for a full-game run, and yes that is the correct name of the level. For the most part the level went well. I got the best shutter cycle at the start with a combination of good movement and a trick to fall through two shutters at once. Before the large room I pause to switch controller. Normally I wouldn't pause for a controller switch because it costs real time, but this is a heavily cycle-based level, and pausing means I won't miss a cycle on account of the switch. Unfortunately the clip went less than perfectly, which ultimately cost me a cycle and 8 seconds. I decided to shake it off and keep going though, because in my opinion it's the hardest of the consistent clips in the game due to the crazy lag in that room, and it was the only real mistake. The end went well, with a nicely done "RNG" boss skip! (Hint: It's not RNG.)

You might notice the camera changes for me in a different order on this level. This is because when you're near a ceiling and try to change from 1st person, the game skips the overhead view and goes directly to the behind view. Routing in pulse also makes the sweeper room much smoother, though it's not always possible to keep pulse for this level. Another interesting aspect of Sewers is that the room before the tunnel switch is probably the best example in the game of using the jetpack momentum glitch for something other than clipping.

Japanese is advantaged due fall mechanics differences. In many rooms you don't take fall damage or can't get a falling death; you don't need to jetpack to break your fall and you maximise freefall boost. Higher gravity also means the falling sections are quicker. This level is tied to an 8 second cycle near the end, so if you get there early you have to wait, so this at least means that Japanese is 1 second easier to save 8 seconds. The best cycle is not verified possible on English, so it may simply be an 8 second advantaged level.

Xizor's Palace — 4:31

Excellent time, with only 5 seconds of mistakes in total. Despite taking a lot more damage than normal at the start, I didn't get hit much overall and had plenty of health at the end. The level as a whole went very smoothly with only a few minor mistakes, and most of the time lost was in the Gladiator fight. First phase was a little slow because I had less ammo than usual, I died unavoidably in the second phase (very small time loss in this case), and the disruptor timing generally could have been a little better to get a bit more damage out of them.

The main difference in strategy for this level are the new "stunner strats" in the bomb room (the tall room with the pillar in the middle). In the past you could easily die in this room if you weren't careful, and sometimes had to replenish health with a checkpoint death beforehand. With stunners you can stop most enemies from shooting you, which lets you take the shortest path and makes the level almost risk-free. Stunner strats also came before cutting flamethrower in Suprosa, which means at one point the speedrun of this game used every weapon, which is pretty cool in my opinion.

Japanese is slower for this level mainly due to disruptor doing much less damage. On English versions you can take out each Gladiator phase with two disruptors each; on Japanese two disruptors do at best 75% damage, which slows down the fight a little. Additional time is lost because of this in the level itself because you need to save seekers to speed up the fight (which otherwise would be used to speed up the level). I estimate English to be advantaged by about 5 seconds.

Skyhook Battle — 6:21

This was an excellent time, although for me a little below-average; this is my most consistent level. In the first part I put my controller down for a while, as there is absolutely no risk whatsoever of gaming over, though I shoot a little toward the end to guarantee I have a spare life. The skyhook part itself had great flying overall. I use the roll mechanic to adjust my flight path while boosting as much as possible. I turned away from one turret slightly later than I could have, but other than that the outside went perfectly. Inside the skyhook is a lot tougher to maximise boost, but that went well. The ending cutscene only starts when you're outside the skyhook again, so the strategy is to suicide against the core after it's at 0% to respawn outside and start the cutscene immediately (this is what the extra life is for). I lost a little time because I bounced once before suiciding, but it could have been worse; I reasonably could have saved 1 second on this level.

Skyhook is split up into two parts. The premise of the first part is that it's an autoscroller where the only objective is to survive for 4:45. Killing enemies serves no purpose other than to gain bonus lives, and dying doesn't lose time. In the second part you need to destroy all the turrets to remove the core shield, then destroy the skyhook's core itself. The second part is all about flying efficiently, and the strategy has actually changed dramatically over time to eke out as much efficiency as possible. This is most obvious inside the skyhook, but there is actually an individual level-only strat for the turrets involving missiles to turn away from them earlier.

This level has no known version biases, and like Asteroid, is the second level with no known version differences at all.

Thank you!

Finally, I would like to thank Alaris, Aleckermit, Flickerform, FnBigIndian, Garnitrex, GenisWon, nolife9face, PitIcarus, Striker and all the old-school players for their contributions to this game. Thanks to SDA for hosting runs like these and running the GDQs. Shoutouts to Adarax, Adde, Aki, Amalthes, Audacity7, AUTOc, Balatee, Barhunga, Bluntslide, Brittany, CalebHart, Cheeze, Choco, Daecks, DireWolf, Dromasylvia, Elminster, Elo, Escro, FootBigMike, Ghoti, Gords, Gregortixlkyns, Grokken, Jarvitz, Jbop, Kejsmaster, Kosmicd12, LeishaChu, Lulz, Maxx, McMitcho, Miscelaneous, Moltov, NatalyaHasDied, NickWolf, PhoenixFeather, Pokey, Qweczol, Qwerty1605, Runnerguy2489, Shredwot, SimpleFlips, SlaskPrask, Summoner, SuperJohn2K9, Tedeth, Tokyoboi, Tompa, Vulpone, Whatsoup, WhyIEyesYa, Yashichi, Zas, Zellpree, Zxv and everyone who watches me on my Twitch channel.

If you want to see more of my speedrun shenanigans and related drama then please follow me on Twitch and Twitter and subscribe to me on YouTube, as my ego is directly proportionate to the numbers associated with those accounts. Also, I don't usually play games as terrible as this one, promise.