Before anything, we’d like to give a special thanks to the following people for their help with strats, being TAS pioneers of SA2, or for just being around to help whenever they can:
THC98: I joined this project while itsPersonnal was still optimizing City Escape. We took a very long time to do the ramp abuse skip, and it turned out to be very optimized. We improved a whole second over the TAS made back in 2012, which was a great incentive on the next stages. A few more seconds were improved in Wild Canyon, due to getting a really good emerald layout in the 2nd possible frame. And then it came Prison Lane. This level was very annoying to optimize, but we ended up improving a LOT over 2012 TAS! And in the next levels, we kept getting times that we could never imagine that they were possible! We also had a few other annoying levels to optimize, like Route 101 and Pyramid Cave. But we still got through them! So overall, I’m so glad that itsPersonnal asked for help in this TAS and I’m so happy to see this masterpiece finally finished, with a really high optimization level!
itsPersonnal: It’s crazy to think that this is actually finished. I started this as a project with Sonicpacker and Mapler back in 2012. Progress came to a complete stop once we hit Prison Lane (the first Tails level), and after a 2 year hiatus, I tried to start it up again. But new strats were found in City Escape so I had to restart from the beginning. I added all the current SA2 TASers to a skype group and asked for help on the TAS, and THC98 was the only really up for the task. I think for the first few months, the idea of actually finishing never went through my head. I dreaded the idea of TASing Route 101 and some of the Tails levels. And to be honest, I never would have got past those levels on my own. THC98 was a great partner to have. Though we certainly reached sections where we had no idea how to make progress, having someone else around to test ideas and try to improve on is really what got us through to the end. Also the vast majority of “entertainment” during bosses and other sections where you have time to kill, was done by him, which was really nice because I don’t like doing entertainment. This was easily the most incredible TAS project I’ve worked on. We broke so many records, even with the lack of time saving upgrades. We found so many new tricks and techniques for optimizing speed and movement. We pushed Sonic Adventure 2: Battle to new limits. And I’m so excited for the public to finally see our finished project.
About this TAS:
The rerecords count is just an average, because the revision of Dolphin that was used doesn't show an accurate number of rerecords
Completes Hero Story
Aims for real time over in-game time, although we tried to optimize in-game time as much as possible without sacrificing real-time. IL tricks like stopping time were ignored in this TAS.
Manipulates RNG for optimal hunting layouts
Uses soft resets to save time
Abuses glitches
GOTTA GO FAST
Final time:
According to the VI count (without loading times): 25:11.87
According to the frames count (with loading times): 27:12.38
According to the in-game time (sum of all levels and bosses): 18:09.48
Commonly used techniques as Sonic:
Bouncing
Bouncing is a game breaking trick that is done by pressing B 30-20 times/sec (depending on the terrain) and holding sporadic directions. You can build speeds up to "17" incredibly fast. Where as a fully charged spindash caps at around 10 speed.
Spindashing
Prior to collecting the bounce upgrade, spindashing is commonly the fastest way of movement. Essentially you just chain spindashes.
Storage/Instant Spindash
Interrupting Sonic while he is about to start a spindash will store his state, allowing you to spindash instantly when you go back to an "on the ground" state. This is often done using springs to interrupt Sonic.
Ramp Abuse
There are two kind of ramp abuse, one during the City Escape boarding section and one on any other kind of ramp. On the boarding section, jumping right before hitting the ramp and then jumping again while on it, this will give you significantly longer air-time. Going at the ramp sideways gives even more air-time. When in a normal state, Sonic can jump off ramps on a skewed angle by jumping before landing on a ramp and then jumping again.
Loop Speed
A technique that was reinvented during this run. Loop speed is performed by taking advantage of scripted loops weird properties. Essentially you just need to homing attack while inside a scripted zone extending out the back of a loop.
Grinding
Some rails have speed caps that are lower than Sonic's speed when he gets on them. You can lock your unexpectedly high speeds by putting Sonic into a "falling off the side of the rail" state and NOT holding B.
Gravity Control/Wallruns
If sonic is moving fast enough off an angled surface, his gravity angle can be skewed temporarily, allowing things often too steep to be considered ground, well, to be considered ground.
Super Bounce
Super bounces convert forward speed into vertical speed by bouncing right before hitting a wall with a high speed.
Mock Super Bounce
Spindashing into certain objects at high speeds can sometimes give an effect similar to a super bounce. It can also be done by spindashing off a ledge and hitting a steep angled surface. This trick was used in Metal Harbor and Pyramid Cave.
Commonly used techniques as Knuckles:
Abusing Down Slopes
Drill diving onto something that is downhill relative to Knuckles’ gravity amplifies speed, like a lot.
Glide Breaking
Glide breaking is a technique used to slow your descent while gliding. When you initially start a glide, you're falling speed stays relatively low for ~60 frames, and then starts to accelerate. By doing a 1-frame drill dive and then starting a new glide, you reset that ~60 frame timer and can cover far distances without losing much height.
Rocket Abuse
Rockets can give weird gravity angles if you drill dive as soon as possible after getting off them. The skewed gravity can be used as a massive speed boost.
RNG Manipulation for emerald layouts
By skipping the cutscene before a hunting level in later frames, we get different emerald layouts. So we simply wait for an optimal emerald layout to come before doing the stage itself. This is the only way to manipulate RNG due to a soft reset right before the stages.
Torch Clip
By jumping below and right in the center of a torch, Knuckles grabs the wall and gets pushed out of bounds by the torch.
Commonly used techniques as Tails:
Pause Jumping
Abusing Pauses to press A twice in two in-game frames will give Tails a height boost that normal hover jumps can not do. Some tricks and out of bounds can only be done with a well timed pause jump. (Holding Y and X makes the pause screen invisible, so that’s why you never see it in the run).
Vulcan Cannon
The Vulcan Cannon shot travels faster than the Laser Cannon and also shoots immediately after pressing B, instead of needing to lock on. With perfect facing angles, you can snipe enemies from across rooms (or even through walls) with the Vulcan Cannon to save time.
Jumping as little as possible
Tails moves fastest on the ground, minimizing air-time saves more than you might think.
Turn around to get speed
When Tails’ speed is very low (or if he’s stopped), it’s useful to do a turn around by holding the opposite direction as he’s facing to. When the turn around animation ends, Tails gets the max speed allowed by the speed cap in the ground he’s on.
Individual levels comments:
City Escape (1.06.50)
We saved an entire second off the 2012 TAS on this stage. Mostly due to more knowledge of how to optimize the boarding section, and the trick to get onto the rail in the section after the giant skip. I think this level is very very close to being capped. More-so than any other stage, this one seems basically maxed.
Big Foot (0.21.39)
All hits were first frame. Nothing special here.
Wild Canyon (0.06.95)
This is the best emerald layout ever, we could never imagine this! Some down slopes were abused to make this time possible.
Eggman 1 (0.11.20)
Dies to stop the timer early. All hits first frame.
Prison Lane (1.21.70)
This huge improvement from the last TAS of this stage (that got 1:27.65) just came from overall optimization, believe it or not. Some jumps were avoided and the movement got much more smooth.
Metal Harbor (1.10.62)
Here is where it starts getting interesting. In the first loop, a wall was hit while Sonic’s gravity was facing sideways so he could get bounced from it. This trick was repeated in another loop in the stage to save even more time. Right after the first check, a loop speed trick was done and it allowed us to do a crazy wall run that saved a lot of time.
Shadow 1 (0.10.94)
All hits were first frame. Nothing special here.
Green Forest (0.58.47)
So much loop speed! This time is unbelievable, it beats the old IL record that had the bounce upgrade! And everything due to the loop speed trick.
Pumpkin Hill (0.19.07)
It’s possible to skip the upgrade cutscene by making a hint monitor text to show up while the cutscene is playing. We just have to activate the monitor before getting the upgrade. About the layout, the emeralds were pretty well placed over the stage, so it was pretty good.
Mission Street (2.00.69)
Cycles make this stage pretty boring normally, but they actually made the first check interesting. We had 2 frames of leeway to make early cycle before check 1, and it costs about 6 seconds to get second cycle here. This was so tight because of how the platforms still move during the upgrade acquiring cutscene.
Aquatic Mine (0.08.67)
This layout was incredible. It was over 3 seconds faster than the second best layout we found! Digging up the last piece was near-pixel perfect to avoid hitting water but still landing inbounds.
Route 101 (2.12.17)
Drifting locks your speed on flat ground.
Hidden Base (1.30.29)
Very nice new strats came from doing more pause jumps.
Pyramid Cave (0.48.34)
Ridiculous first check time, and cool strats up until the bounce upgrade. Everything from there on you just need to see for yourself.
Death Chamber (0.18.47)
This doesn’t look like a good layout at first. But because of the many out of bounds tricks that can be done here, this one was insanely good. A lot of RNG manipulation was done to get those speed shoes. It was manipulated by digging in slightly different spots until speed shoes were got.
King Boom Boo (0.34.67)
It’s possible to go through King Boom Boo by going in a certain area of its body. But when Knuckles get hit from him, the stick control gets locked for a few seconds, so a spiral upper was done to change the facing angle so hitting the hourglass was possible. This is why there were a lot of punches after getting hit.
Egg Golem (0.18.04)
The first 3 hits were done by doing a super bounce on the fence of the platform after a few bounces. In the last 2 though, as Egg Golem’s arm was close to Sonic, it was possible to bounce on it and save a few frames (and it looks very cool!).
Eternal Engine (1.48.40)
There wasn’t anything special about this stage. Movement was a little hard to optimize, mostly on the big skip though.
Meteor Herd (0.20.67)
No layout could get a so fast and so amazing strat like this one. The wall run after the first piece was simply crazy!
Rouge (0.14.39)
All hits were first frame. Nothing special here.
Crazy Gadget (0.29.82)
AHHHHHHHHHHHHH CRAZY NEW STRATS
Eggman 2 (0.14.00)
All hits were first frame. Nothing special here.
Final Rush (1.02.57)
This is almost identical to the old IL record, but a new strat to skip one of the rails and save time was found. All other time-save came from more bouncing and rail optimization.
Nahoc: amazing audience response! Great TAS! Although it is made on a pretty old dolphin revision, I see no reason to not publish it. Good job guys and good luck with the future "Dark Story"!
Was that the only Eternal Engine clip you were able to make work?
I know there's a tank in a similar position in the very first room, which is at a height of -31, and the goal is at -1798. Horizontally, that endpoint is just over 9000 units away, so the question, beyond does the geometry actually allow a breakout there, is whether hovering allows you to fall slowly enough to travel on a 1:5 slope.
In the Dreamcast version where there are no invisible barriers blocking such a path, it's just barely possible to backtrack on the fly from checkpoint 5 (-1680, +944, -9150) to the corner of the falling weight room below checkpoint 2 (-690, +177, -4790). A perfect hover jump gives an extra 50 height from the departure point, but even an unassisted 35-height jump can make it, for a distance of 802 vertical units and 4471 orthogonal units, easily better than the required 1:5.
You sped past the last section of the run quickly enough that Eternal Engine was never part of a WIP to give feedback, but if that wall in the first room is breakable, then you can most definitely hover in a straight line start-to-finish. If not, you might also consider trying to clip into the exploding canister tube in the corner of the room behind checkpoint 2, which is easy to do on DC, and yields a route similar to what you already have minus the sidetrack to the left.
Was that the only Eternal Engine clip you were able to make work?
I know there's a tank in a similar position in the very first room, which is at a height of -31, and the goal is at -1798. Horizontally, that endpoint is just over 9000 units away, so the question, beyond does the geometry actually allow a breakout there, is whether hovering allows you to fall slowly enough to travel on a 1:5 slope.
In the Dreamcast version where there are no invisible barriers blocking such a path, it's just barely possible to backtrack on the fly from checkpoint 5 (-1680, +944, -9150) to the corner of the falling weight room below checkpoint 2 (-690, +177, -4790). A perfect hover jump gives an extra 50 height from the departure point, but even an unassisted 35-height jump can make it, for a distance of 802 vertical units and 4471 orthogonal units, easily better than the required 1:5.
You sped past the last section of the run quickly enough that Eternal Engine was never part of a WIP to give feedback, but if that wall in the first room is breakable, then you can most definitely hover in a straight line start-to-finish. If not, you might also consider trying to clip into the exploding canister tube in the corner of the room behind checkpoint 2, which is easy to do on DC, and yields a route similar to what you already have minus the sidetrack to the left.
The tank in the very first room isn't actually in a good position to make the clip be possible, as far as I know it's only possible to clip through those tanks if they're in the corner of a room.
About the exploding canister tube clip, I haven't got it to work on GC version. And I don't know either if there are kill planes around there (there were some of them in the spot we clipped). Can you show us a video of that clip on DC or tell us how it's done?
Thanks for the idea though :)
EDIT: About the emblem counter showing 9 emblems instead of 17, itsPersonnal has a more probably idea:
My guess is actually the "kart racing unlocked" text that plays after Route 101 (the 9th emblem) is the reason it stops at nine. Maybe someday I'll do a little test and collect an extra emblem before Route 101, and see if it jumps to 10 after.
Absolutely loved this! Makes me want to hunt down a copy and play this... that and the soundtrack was a lot of fun! [Not enough games make you want to find a copy of the soundtrack to enjoy outside of the game!]
Fast, furious, fun! Yes Vote!
Mr. Kelly R. Flewin
Mr. Kelly R. Flewin
Just another random gamer
----
<OmnipotentEntity> How do you people get bored in the span of 10 seconds? Worst ADD ever.
Great work, the both of you. I was looking forward for sure, but wasn't super hype'd for this... just as I've been following the IL's for the game for ages, so wasn't expecting too many surprises. Well you proved me wrong.
Looks like that Il Spreadsheet of yours ItsPersonnal will need redoing at some point huh! And looking forward to the next game you move onto THC, whatever it is.
Fantastic stuff.
My god. I lost my shit when you got the bounce bracelet.
This is amazing.
Hopefully a dark story is being done as well because I couldnt handle not seeing dark story broken too.
Joined: 8/8/2005
Posts: 296
Location: NSW, Australia
I knew that King Boom Boo method was possible I JUST KNEW IT
Been waiting for over a decade for this to finally be made and show just what this game was capable of. Pyramid Cave was ballistic.
Easiest yes of 2015. Calling it early!
Joined: 9/21/2009
Posts: 1047
Location: California
GHNeko wrote:
Hopefully a dark story is being done as well because I couldnt handle not seeing dark story broken too.
This run is long defunct, but a cool preview considering it's only level 1! ;)
On that note, yes vote, and I'm glad I found the softreset to be faster after levels. It makes the fluidity of this run even more apparent.
I'd love to see a full hero, Dark, and secret run in one segment. and getting all A in all stages (12345) missions as well. a true 100% run.
180 emblem or RIOT
I second this!
pS:But I have a suggestion for the 2 other stories: Do a Dark story starting from the file of the Hero story. Publish it as a Dark story. Then, a movie that completes both story and the Last story(The one where every guy is played one time). This movie will obsolete the two others movies?
Since it uses a lot of reset, i figure it won't be hard to combine both stories... I think?
Joined: 9/21/2009
Posts: 1047
Location: California
niamek wrote:
Do a Dark story starting from the file of the Hero story. Publish it as a Dark story. Then, a movie that completes both story and the Last story(The one where every guy is played one time). This movie will obsolete the two others movies?
This was the original plan. Minus publishing Dark Story. Anyway, all you have to do is soft reset again and go to Dark I assume.
Do a Dark story starting from the file of the Hero story. Publish it as a Dark story. Then, a movie that completes both story and the Last story(The one where every guy is played one time). This movie will obsolete the two others movies?
This was the original plan. Minus publishing Dark Story. Anyway, all you have to do is soft reset again and go to Dark I assume.[/quote
Yeah it's true, last story is essentially one level only.
[8.8%] 8833/100528 frames, 0.25 fps, 3444.22 kb/s, eta 100:57:23
UPDATE :D
[68.8%] 69126/100528 frames, 0.22 fps, 3850.27 kb/s, eta 38:55:32
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
I'm pretty sure it's an internal problem: Dolphin is missing samples every once in a while, we force it do add gaps there, then these gaps are fixed not to sound crackling, the result is what you hear. Like, anything we could do with it would sound worse, or desynced.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
I'm getting an offset of 2 seconds on the audio, from the beginning.
"The guy was fatally injured and wants to be covered by God's tears (rain) before he dies. God is too busy to bother because it wastes frames."
Frames 16:26
HQ Archive.org. Sorry, I forgot to specify.
Setting MPC to use an audio offset of ±2000ms fixed it. Usually "laggy" issues are variable and setting a solid offset wouldn't help.
"The guy was fatally injured and wants to be covered by God's tears (rain) before he dies. God is too busy to bother because it wastes frames."
Frames 16:26
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
At least we haven't started console verifying handheld games yet. I wonder what decides if a post is to short. And I hope it isn't some sort of attempt at ensuring post content is meaningful because we all know verboseness does not equate to meaningful content.
"The guy was fatally injured and wants to be covered by God's tears (rain) before he dies. God is too busy to bother because it wastes frames."
Frames 16:26
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
