Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
In my hack, go to options, game options and change coop mode from 'classic' to 'chaotix'. It adds Knuckles' Chaotix-style energy bands that pulls the players to one another.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Nice work. Some things will not work so smoothly when I finally finish porting the S3&K camera system (because it moves faster), but this TAS should survive transition to the next version.
By the way, you do know you can just ask about the RAM addresses, right?
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
The glitch death in HCZ1 requires you to zip out of the screen far enough so that the object that lets you walk on water gets deleted. Entering a special or bonus stage causes this object to be recreated as the level loads. Thus, you can't get an emerald and perform it unless you zip in between.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
TheYogWog wrote:
How come you only have to redo AIZ and not the rest of the levels? Is it easy to resynch them after that point, and is it likely they will stay in synch up to SOZ after that? Basically, how does resynching work?
Resynching the other levels is easy once you know what to do; I suffered a lot more to resynch levels in the past because I did not understand all factors involved. There are two basic sources of desynch in this game (or most games, to be honest): lag and luck.
Sometimes, lag can push around the input at the start of the level; this is easy to resynch using TASEditor. Sometimes, there is more (or less) lag inside a level; this is a lot more annoying to resynch, but can also be done with TASEditor -- you compare both runs near the point where desynch happens and add add or subtract frames to compensate for lag difference. Changes in lag between levels is very rare (it is much more common in S1, a lot less common in S2 and S3&K).
Luck differences are due to differences in the V-Int run count (i.e., number of frames since power on modulo 2^32) having different values in "important" points -- say, when it is used to determine whether or not to use the RNG. This is more annoying to manage, but in many cases, it can be managed by manipulating luck -- adding pauses in strategic points can cause some objects that would sample the RNG to either never be created (keeping the RNG seed intact for some more time) or to be created faster than they otherwise would (rapid-fire changing the RNG seed) so the RNG seed can have a value you want at a frame when it is important. In other cases, you just have to add enough pauses (or remove them if they were there already) because an object uses the V-Int run count to determine when to toggle its collision (say, the thrusters of the MHZ1 mini-boss) instead of sampling the RNG.
The most important thing that level wraps and zips do is remove randomness, so luck becomes less of a factor. I anticipate needing to resynch MGZ due to lag, CNZ2 due to luck -- those damned balloons always get in the way (done by deleting sparkles from the act 1 signpost), MHZ1 boss due to luck (by adding or removing pauses at the start of the act to have his exhaust flames hit Tails when I need them to) and MHZ2 due to lag.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Hehe, it is entertaining.
Now, some bad(?) news: I just realized that there is a 11-frame improvement to HS+T AIZ1. This means redoing the level, redoing AIZ2 and then resynching the run up to SOZ again...
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Ah, so qwerty has finally figured out you can drop down levels in MGZ1 for the level wrap.
Anyway, due to the airline cancelling my flight, I could not return home until now. You will be seeing the WIP of the newgame+ run tomorrow.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Let's see: for starters, there is not a "the" time axis, just as there is no "the" x axis — things/space do not travel around this non-existent time axis anymore than things/yz planes do around "the" x axis. The choice of a time axis, like the choice of a x axis, is a choice made by the observer, but with some constraints — it must be a timelike vector — which I won't go into right now. In fact, you can work on coordinate systems that don't have a time axis without any issues.
The truth is that a notion of time is established locally by the light cone, and you can extend this to a region if space-time is time-orientable in that region — most physically relevant cases are. Things travel in THEIR time axis, which is defined by their 4-velocity vector (which does not depend on observers or coordinate choices). Likewise, "space" — understood as being a hypersurface of constant time* in some coordinate system — bends and twists according to the local to the local time axis at each point.
Finally: yes, moving in a more "curved" time axis changes how "far" you travel — how much proper time elapses between the endpoints. The hyperbolic nature of space-time means that, contrary to Euclidean expectations, you travel less in more curved paths through space-time than if you travel along the shortest (geodesic) path.
* Technically, it is a "spacelike hypersurface".
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
For what is worth, this particular improvement I suggested is the same one that is used in the new Hyper Sonic run (it has been there for quite some time); you will be seeing it tomorrow, when I am back home. Normal Sonic can use this particular trick only with a lightning shield, so don't be surprised when Aglar posts the AIZ1 video and it isn't there.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
For what is worth, it was only me that asked that WST keep it under wraps — but Aglar has talked to me and we will be releasing AIZ1 with the trick so WST is free to show his work.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
The caps lock is used for some other hack, I don't remember right now which one. Changing keybindings should have no effect on synch, they are not stored in savestates or movies; would your setup let you bind to the same keys with numlock off or are the underlying keys (arrows, page up/down, home, end) used for other purposes?
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
For what is worth, the last debug glitch I investigated (spawning the Marble Garden 2 boss in Hydrocity 1 due to death + debug move) ended up throwing address errors on Regen. Neither Gens and Kega Fusion emulate address errors, but they sometimes have effects other than locking upresetting the game in these emulators due to executing invalid instructions: one of the forms they manifest in Gens and Kega is graphical glitches similar to what appears in the video. To be sure, one would need to try to replicate these tricks on an emulator such as Regen, Exodus or Gens Plus GX; if you only ever get lockupsresets on these emulators, then the debug glitches would be due to an (intentional) emulation error.
Edit: I also haven't investigated the Marble Garden glitches with the spinning top, but it may well happen to be the same thing.
Edit 2: I had forgotten, S3&K resets instead of locking up on crashes.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
The hidden bonuses always count when you get them; what really happens is that the 100 point bonuses are bugged and give only 10 points instead (this also happens in the original S1).
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
For reference, HHS is absolutely correct -- potential energy is as real as kinetic energy, or chemical energy, or whatever other form of energy you want. Except for New Age bullshit "energy", but that is not relevant here.
But gist of the issue is that "potential energy" is a highly misleading name: there is a load of things that are very real that get assigned that name, such as electromagnetic potential energy, gravitational potential energy, and so on. These are only "potential" because we misleadingly call them that -- they are very real interactions of the objects we are studying with well defined and measurable fields. Where does the "kinetic energy" go when you climb up a gravitational field? Into the gravitational field. Where does the kinetic energy come from when you fall back? From the field. Where is the elastic potential energy stored in a spring? In the electromagnetic interaction between the atoms of the spring. And so on. The energy is real, and it is all there; it is so real, in fact, that general relativity says right out of the box that it generates gravitational fields. I can elaborate on that, if so desired, but I won't do it now.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Well, as it happened before, so it happens again: Kiske sent me an improvement to LZ1 which I thought was impossible; I further improved his effort by 7 frames, leading to a 92-frame improvement in LZ1. Here is he new movie.
It is mighty annoying; the last WIP stands in the movie thread for a full week before I submit, and AFTER I submit it, people come up with improvements...
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
New version of the script is out, probably the last download that I will add to Google Code (since they are disabling downloads due to abuse).
Changes:
Replaced score display by camera position display.
Added a level bounds display which shows where the player is stopped. Yellow is for player 1, magenta for player 2; magenta is drawn after, so if only magenta is being shown, then both are being coincident. These displays are keyed by the character's HUDs, and will only display if those are also being displayed. The computations in the games for level bounds use the player's centerpoint; the HUD uses their hitbox instead. This means that the boundary will shift when the character's size changes.
Found and added horizontal scroll delay for s1tails and for s1knux.
Fixed boss code tables for S&K and S3&Amy
Build scripts now skips missing ROMs and warns about it.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
It was JXQ, yes.
Anyway, here is the preliminary full run, clocking a total 6:10::24 in-game time, for a total improvement of 6618 frames over JXQ's run.
SBZ1 ended up being a mere 12 frames faster. I started the SBZ2 run based on a version Kiske sent me, but it quickly diverges as I do neither a glide nor a spindash before zipping right. I also glide-glitch through the floor at the transition. SBZ3 uses a trick Kiske sent which has Knuckles fall through the moving platform at the start. Tails can't do this because of his smaller hitbox, but he is still faster in this act because he moves more efficiently inside the wall than Knuckles.
Now to scrub for frames... then manipulate the animals, which will entail editing a pause in LZ1 to manipulate the burrobot.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Just a status update: so far, SBZ1 has resisted every attempt at improvement -- the zip from Sonic's run is 2 seconds slower than JXQ's run for diverse reasons, a level wrap using slope glitch sadly results in death before being able to do anything (as does extending zips all the way to the right abusing glide and zip speeds), and so on. Still have a couple of things to try, but it looks like this level will not cooperate at all.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
New WIP up to SLZ3. This one features better use of slope glitch in SYZ3, leading to the first time that level has been done in under 30 seconds. I am using WST's SLZ1; I don't think it can be done faster than this.
SLZ2 is 4 frames faster than Tails' due to spindash; Tails is ejected up from the wall if spindashing there, while Knuckles' isn't (same happens in SLZ1, actually). The SLZ3 route is basically Tails' but slower; I did manage to improve the boss fight by 10 frames too.
By the end of SLZ3, 4974 frames have been saved. SBZ1 will break 5000 if a few things I am thinking of pan out.
Experienced Forum User, Published Author, Experienced player
(758)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Damn, but this hack is broken; I continued on to LZ and so far, Knuckles has been steamrolling through every level; check this.
This has the new GHZ1, the improvement WST suggested for GHZ3 and all levels up to LZ3. While I was testing how to hit the monitor and switch in LZ2, I discovered that roll-jumping from a monitor breaks it and gives you slope glitch in this hack; so I finally found a way to break SYZ3. This hasn't been useful in any other levels so far; it likely won't be in SLZ, SBZ2-3 and FZ, but it can possibly be useful in SBZ1 -- I haven't tested yet.