Started at the same time as the Tails run, and worked on at the same time as its twin run, this run has been years in the making. Like the Tails run, a disassembly of the game was used in order to squeeze the most of every trick. This run saves 03:00.45 in real time over its predecessor, 02:43.81 of which is saved in in-game time. Every single level has been improved from the previous run, and none of Nitsuja's or Upthorn's input survived.
Note that the rerecord count is an estimate.
Emulator used: Gens 11 svn 351 + S3KCamhack + SolidityViewer + (partial)HitboxDisplay + LUA HUD for Genesis Sonic games

Game objectives

  • Ignores delays caused by bonus effects
  • Aims for fastest (primarily in-game) time
  • Takes Damage to save time
  • Abuses death
  • Abuses programming errors
  • Counts time spent dying
  • Manipulates luck
  • Has one speed/entertainment trade-off.

Comments

General Comments

Note: Comments by Marzo.
This run was made with constant reference to the disassembly available at Sonic Retro. This disassembly was used to determine the exact positions where the camera locks, how bosses and badniks behave, how several glitches work, and so forth. All this information is being collected and revised, and will be released in the future.

In-Game Time Table

ActTime[1]Previous[1]Change
Angel Island 10:35::550:38::45-170
Angel Island 21:14::531:16::23-90
Hydrocity 10:18::370:30::26-709
Hydrocity 20:21::580:23::37-99
Marble Garden 10:50::560:29::291287
Marble Garden 20:03::240:35::36-1932
Carnival Night 10:32::280:34::50-142
Carnival Night 20:40::220:41::41-79
IceCap 10:59::061:00::14-68
IceCap 20:00::430:00::44-1
Launch Base 10:27::520:31::55-243
Launch Base 20:25::500:26::26-36
Mushroom Hill 10:32::270:32::50-23
Mushroom Hill 20:32::500:34::05-75
Flying Battery 10:42::490:56::16-807
Flying Battery 21:06::131:11::21-308
Sandopolis 11:10::131:39::16-1743
Sandopolis 20:57::191:09::47-748
Lava Reef 10:48::481:03::23-875
Lava Reef 20:20::530:21::40-47
Hidden Palace0:16::360:17::27-51
Sky Sanctuary0:52::130:52::25-12
Death Egg 10:57::281:04::03-395
Death Egg 21:37::152:18::06-2451
Total17:02::0519:45::41-9816

In-game time lost to deaths

ActTime[1]Previous[1]Change
Hydrocity 20:05::210:05::210

Cut-scenes

ActTime[1]Previous[1]Change
Launch Base 20:29::360:29::351
[1] The value after the final colon pair indicates the in-game frame counter, and is restricted to the 0-59 range. It is updated every frame (except during lag frames), and increments the in-game second when resetting to zero.

Level-by-Levels comments

Here we mostly only mention what's not obviously clear when watching the run, each other situation pretty much speaks for itself or has been used in earlier TASes of this game.

Angel Island 1

I had already improved the start a lot; Aglar went in and took it even further. It is possible to perform the loop zip on the second loop too, it is useless without a lightning shield; Knuckles or Hyper Sonic can also make it. Delaying a bit after the scenery burns down lets me bounce off the spring on the top and lets me carry on more speed

Angel Island 2

Improved precision and a different route to the switch.

Hydrocity 1

A 2-player variation of the stair clip is used here, which in this case was much harder to execute as the object that stops the horizontal movement resides quite far from the edge of the terrain. The quick death for the next act was setup at no cost thanks to Tails.

Hydrocity 2

Better camera management for the level wrap, plus a new way of getting into the boss arena that allows a frame-perfect kill.

Marble Garden

Tails moves as in the Tails run, carries Sonic along. When landing, Tails touches down 1 frame before Sonic, so the flight timer is not set and allows Tails to pickup Sonic right away later. First lightning shield is skipped, the second one is picked up with no cost in time. Improved precision carries the rest, with a surprise twist: Nineko's trick with solo Sonic or Knuckles of doing both acts at once can actually be done as Sonic + Tails. The condition for it to work is that Tails must be off-screen at one specific frame; this is not normally possible if you don't know the condition, which was figured out through the disassembly. The pattern at score tally minimizes the number of lag frames.
By the way, I was a bit sad about finding this trick: it is possible to use the route from the Tails TAS in act 2 and hit the boss above the ceiling; Sonic and Tails can deliver all 8 hits this way, but it ends up being a lot slower.

Carnival Night 1

Scrolling a barrel and a set of spikes off-screen, then being dropped through them by Tails leads to a faster level wrap. Entering the wall at the boss allows zipping out at at the right time, allowing for a nearly frame-perfect kill -- it is theoretically possible to beat the boss 4 frames faster, but Sonic simply can't move fast enough for it. The original version of the ground entrance was by DMTM and used a bubble shield; I proved it was possible to do it without the shield.

Carnival Night 2

Zip from the start, better precision and better camera management.

IceCap

Better precision all the way. 43 frames of "act 2" after the boss are now accounted; previous runs did not count this time, which made them 44 frames shorter in in-game time than they actually were.

Launch Base 1

Based on the Tails trick and on the Mushroom Hill 1 door, this trick is almost as old as the version with Tails. Better boss fight, together with extra setup for death right after signpost stops, allows skipping the score tally entirely.

Launch Base 2

Better boss fight from lodging Sonic inside the pipe. Unfortunately, it is not possible to speed up the transition, and it got longer by 1 frame.
Warning: If you watch this level with a camera hack, be sure to get the latest version of the S3KCamhack in the thread, or Gens will hang.

Mushroom Hill 1

It is theoretically possible to use Tails to enter terrain right at the start of the level, before the Knuckles cut-scene, and level wrap right away; but this does not work, and Sonic is teleported back to watch the cut-scene. And while he will still be on terrain and be able to zip right away, it ends up being slower overall. Anyway: improved ground entrance.
Warning: If you watch this level with a camera hack, be sure to get the latest version of the S3KCamhack in the thread, or Gens will hang.

Mushroom Hill 2

Improved precision, better management of Tails.

Flying Battery 1

Tee-N-Tee's new zip for Tails can be used here, but it is slightly slower than this.

Flying Battery 2

Used the boss of the previous act to gain slope glitch, and bounced the signpost to obtain a lightning shield. The latter is crucial for this route, as Tails cannot follow Sonic on this route: CPU Tails despawns when off-screen with slope glitch. The speed of climbing at the end is irrelevant as long as you don't die.

Sandopolis 1

Yes, as it turns out, this zip is possible for Sonic -- it requires entering it at perfect pixel position with near zero speed. I had previously tried only with positions attainable within the wall (trying all subpixels), and thought it was impossible; Aglar proved me wrong. The needed flame shield was nabbed after the boss, and used to setup object positions on next act.

Sandopolis 2

Several new ground entrances, and improved precision throughout. One frame is wasted between two hits on the boss to save 30 from glitching his invulnerability counter.

Lava Reef 1

The Tails route can't be followed by Sonic, so we had to improvise a bit. The two characters can then team up to descend faster and setup a zip left, which locks the screen further left for the cut-scene. The idea of going through the drill came about by watching Amy do it in WST's TAS.

Lava Reef 2

Level wrap is now perfect: using spindashes to keep camera as far right as possible while zipping left, then using a flame dash to nullify the camera lock right after the level wrap.

Hidden Palace

Improved precision up to the teleporter. It is possible to gain 1 frame more up to it, but the teleporter is the one object in S3&K that has a 4-frame frame-rule. Glitched the boss even more, so Knuckles is down before his music starts... and then broke the level.

Sky Sanctuary

Replayed the level entirely, and managed to gain 3 cycles of the teleporter's 4-frame frame-rule.

Death Egg 1

9 frames were gained in the first frame of input in the level. The Tails zip at the start seems impossible; so improved precision and a couple new zips carried the day. Ditched the lightning shield for invulnerability frames, in order to get a frame-perfect kill at the boss. Used Orkal's trick to break the level transition and die with the timer stopped.

Death Egg 2

Level wrap by gamble5688, with lightning shield by DMTM.

Other comments

Thanks goes to Nitsuja and Upthorn for their Sonic runs;
to Orkal for his glitch videos;
to HHS for his explanation of the Angel Island/Hydrocity transition glitches;
to mike89 for his blooper videos;
to HDL for his general awesomeness at this game;
to the folks at SonicRetro involved in the making of the S3&K disassembly.

Potential improvements

There is a small pause at the end of Sky Sanctuary to manipulate Red Eye; if he were to be replayed, as well Death Ball in act 2, this pause could be removed to save 3 frames.

feos: Accepting as a (notable) improvement over the existing run. To whoever would publish it: this run would need a camhacked copy of each encode (WFIW).
Spikestuff: ...SEGA...

Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Sappharad wrote:
But in the case of the run that this thread is about, you don't see any blinking at all. When Sonic is fighting any of the bosses, nothing happens when Robotnik is hit. You should see him flashing, or at least light up if deblinked. At least for me, it looks like nothing happens at all when he's hit.
It's because it was not deblinked at all. Watch my newgame+ encodes for how it should look like when done right. The boss will look light gray, just like in Gens, just like on TV. If I uploaded 60 fps of that boss, he would indeed be flashing, which (as I said) is different from console, Gens, and devs' intend.
Aktan wrote:
What I see in the second video with 1080p@60FPS is 30Hz flicker like the source probably is.
If it's 30 fps blinking, you wouldn't see it. You'd see translucent James. But if you really see how he blinks, then it's indeed what I was talking about. HOWEVER, I have an idea. If we upload 60 fps deblinked footage, it will improve scrolling and still do the translucent job! What do you think? Now we only need to prove that it improves scrolling :D
Aktan wrote:
feos wrote:
I see no proof of that in my version of reality.
Then it could be he was being sarcastic...
Versions of reality do exist.
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.
Spikestuff
They/Them
Editor, Publisher, Expert player (2312)
Joined: 10/12/2011
Posts: 6342
Location: The land down under.
Patashu wrote:
-There are some other weird glitches, such as you can upload in 1080p 60fps but youtube will make the max quality 720p 60fps anyway.
Expanding on this for Chrome users. When uploading a 720p60fps on Firefox you get told it can be played in 720p max (obviously playing in 30fps on Firefox). When you're using Chrome on the other hand the max shown is 480p60fps. 480p60fps is actually 720p60fps. Going downwards from 480p60fps is: 360p30fps which is 480p30fps, 240p30fps which is 360p30fps and 144p15fps which is 144p15fps.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account. Something better for yourself and also others.
Player (144)
Joined: 7/16/2009
Posts: 686
feos wrote:
It's because it was not deblinked at all. Watch my newgame+ encodes for how it should look like when done right. The boss will look light gray, just like in Gens, just like on TV. If I uploaded 60 fps of that boss, he would indeed be flashing, which (as I said) is different from console, Gens, and devs' intend.
30 Hz blinking is exactly what the console does. If you're deblinking it, you're making the footage less accurate. Sure, it looks like the character lights up, but that holds for 60 fps blinking footage too. Deblinking has only ever been done because it's a way to circumvent one of the issues that YouTube's 30 fps limit brings along. With the prospect of proper 60 fps video, we should get rid of deblinking and never think of it again.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Scepheo wrote:
30 Hz blinking is exactly what the console does. If you're deblinking it, you're making the footage less accurate. Sure, it looks like the character lights up, but that holds for 60 fps blinking footage too. Deblinking has only ever been done because it's a way to circumvent one of the issues that YouTube's 30 fps limit brings along. With the prospect of proper 60 fps video, we should get rid of deblinking and never think of it again.
If on TV you don't see what I see after deblinking, I don't know what to say to that. If on YT@60 fps you see the same thing as on TV, I don't know what to say to that.
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
It's because it was not deblinked at all. Watch my newgame+ encodes for how it should look like when done right. The boss will look light gray, just like in Gens, just like on TV. If I uploaded 60 fps of that boss, he would indeed be flashing, which (as I said) is different from console, Gens, and devs' intend.
Aktan wrote:
What I see in the second video with 1080p@60FPS is 30Hz flicker like the source probably is.
If it's 30 fps blinking, you wouldn't see it. You'd see translucent James. But if you really see how he blinks, then it's indeed what I was talking about. HOWEVER, I have an idea. If we upload 60 fps deblinked footage, it will improve scrolling and still do the translucent job! What do you think? Now we only need to prove that it improves scrolling :D
Are you saying that the capture video is actually wrong from Gens and that deblinking fixes that? I guess I need to do more research on deblinking, but I had assume that deblink blurs the background, which I doubt is what the "dev" intended. I could be wrong which is why I need to read more on it. Now as for intended on TV, you could also say scanlines were on the TV which means our encodes should have scanlines... but that's another can of worms, lol.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Aktan wrote:
Are you saying that the capture video is actually wrong from Gens and that deblinking fixes that?
What is that about? If about my first paragraph, then it's the problem I'm describing for quite a few months already: ~60 fps video dump recorded from emulator that contains alternate frame flickering will become 30 fps video with dropped frames when youtubed at 60 fps. It's how it works on their side.
Aktan wrote:
I guess I need to do more research on deblinking, but I had assume that deblink blurs the background, which I doubt is what the "dev" intended.
Wow. Deblink calculates only the object that blinks at 30 fps and makes it translucent. http://tasvideos.org/Nanogyth.html
Aktan wrote:
Now as for intended on TV, you could also say scanlines were on the TV which means our encodes should have scanlines... but that's another can of worms, lol.
Is it really that hard to compare what you see?
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
It's because it was not deblinked at all. Watch my newgame+ encodes for how it should look like when done right. The boss will look light gray, just like in Gens, just like on TV. If I uploaded 60 fps of that boss, he would indeed be flashing, which (as I said) is different from console, Gens, and devs' intend.
This is what made me think that the Gens output is incorrect since it sounded like you needed deblink to fix it.
feos wrote:
Wow. Deblink calculates only the object that blinks at 30 fps and makes it translucent.
That's what I'm reading so far, which is nice, though it doesn't fix the less fluid background on scrolling which is why you suggested deblinking on 60 FPS.
feos wrote:
Is it really that hard to compare what you see?
On what now? I already told you what I saw, which is 30 hz flicker, which could be the same thing as the transparency that you seem to think on TV. What else do you want me to compare? lol.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Aktan wrote:
I saw 30 hz flicker, which could be the same thing as the transparency that you seem to think on TV.
That's why I asked to specify. You see James translucent or blinking here?
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.
Publisher
Joined: 4/23/2009
Posts: 1283
as I said I see blinking, maybe because I know the translucent is the result of blinking really fast but I see blinking. Maybe the confusion is when I said 30 Hz blinking as maybe I'm using the term wrong, but when I say 30 Hz blinking I mean the full blink cycle is completed 30 times a second, aka you need 60 FPS to see 30 cycles of blinking.
AntyMew
It/Its
Encoder, Player (35)
Joined: 10/22/2014
Posts: 425
feos wrote:
Aktan wrote:
I saw 30 hz flicker, which could be the same thing as the transparency that you seem to think on TV.
That's why I asked to specify. You see James translucent or blinking here?
On my side he's invisible in 360p, and blinking in 720p
Just a Mew! 〜 It/She ΘΔ 〜
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Aktan: Yeah, I also should have said "alternate frame blinking @ 60 fps" (source) versus "semi-alternate frame blinking @ 30 fps" (result). So now we know the exact pattern they use, we know what people see on YT under it's feature to imitate 60 fps footage, the only question is shall we start deblinking to 60 fps. Will test tomorrow.
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
Aktan: Yeah, I also should have said "alternate frame blinking @ 60 fps" (source) versus "semi-alternate frame blinking @ 30 fps" (result). So now we know the exact pattern they use, we know what people see on YT under it's feature to imitate 60 fps footage, the only question is shall we start deblinking to 60 fps. Will test tomorrow.
Well I see "alternate frame blinking @ 60 fps" like source at 1080p60FPS. And yes I see "semi-alternate frame blinking @ 30 fps" at 480p from same video. Aka what I see in terms of blinking is different for 480p and 1080p60fps with the later a lot faster. I should note again, if you are unsure if it is playing at 60 FPS, right click the video and go to "stats for nerds" and then read the stats. Also note to see if frames are being dropped. I have no frames dropped and playing at 60 fps when I see "alternate frame blinking @ 60 fps"
Player (144)
Joined: 7/16/2009
Posts: 686
feos, in the video you linked, when playing it at 60 fps, the player is blinking @ 60 fps. If to you, he seems translucent, try pausing: he should either be solid or completely gone.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
I'm absolutely confused then. Because in the source he is translucent! And no one here says he is translucent on youtube.
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
I'm absolutely confused then. Because in the source he is translucent! And no one here says he is translucent on youtube.
Are you saying if you open the source file in VDub and go frame by frame, he is translucent? If that's so, I'm surprise YouTube changed the video that much.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
feos starts puling his hair out... When I replay the source in real time, I see translucent. When I watch it on YT, not matter what I do, I don't see translucent, only blinking a naked eye can see.
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
feos starts puling his hair out... When I replay the source in real time, I see translucent. When I watch it on YT, not matter what I do, I don't see translucent, only blinking a naked eye can see.
Have you done what I suggested which is right click the YT video and goto stats for nerds and made sure it says it is playing at 60 FPS and not dropping frames? Edit: Multiple people has said they seen fast blinking to what you probably call translucent, so I have no idea why you said no one has said they seen translucent. Just because you call blinking every other frame at 60 FPS translucent and we call that blinking does not mean it isn't the same.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Yes, it keeps assuring me there's only 1 dropped frame, and it runs at 60, but I freaking see him blink and nothing more. And you see him translucent, right?
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
Yes, it keeps assuring me there's only 1 dropped frame, and it runs at 60, but I freaking see him blink and nothing more. And you see him translucent, right?
Okay, if you compare 480p to 1080p60FPS is the blinking at least faster or the same? Yes I see him translucent.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Visually, absolutely identical, except that at 1080p it lags as hell, and at 720p the blinking is less consistent.
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.
Publisher
Joined: 4/23/2009
Posts: 1283
You can try 720p60FPS. Try replaying the video over and over a few times with the replay button on the bottom left and see if it runs faster the 2nd or 3rd time around. Edit: Also make sure you are on the latest version of Chrome.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Yes, I've done all that. Downloaded Chrome, ran the clip over again, the result is above.
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.
Publisher
Joined: 4/23/2009
Posts: 1283
feos wrote:
Yes, I've done all that. Downloaded Chrome, ran the clip over again, the result is above.
Well only other suggestion is update your GPU drivers so hopefully it be hardware accelerated. Basically your CPU or GPU is too slow to playback at full speed then. Other suggestions include making sure you are plugged into AC if you are on laptop and set the GPU drivers to be performance.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Okay, so it indeed runs at 60 at least for some blessed people. And it's not going to start supporting old troughs like mine, since it simply runs as it can already. Probably it means we'd need a stats poll like this: http://tasvideos.org/forum/viewtopic.php?t=14097 But again, I'd need to do some real testing again, which I didn't have time for today. Will eventually make a poll this week.
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.
Site Admin, Skilled player (1237)
Joined: 4/17/2010
Posts: 11274
Location: RU
Everybody answer the poll please: http://tasvideos.org/forum/viewtopic.php?t=16264
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.