Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
AnS wrote:
No, wait, since the addresses $CA-CB store the "Camera Y", this value is subtracted from the "Real Y" of an object (to get its onscreen Y).
Normally in 4-1 it calculates like this: OnscreenY = RealY - 0.
But after the glitch it becomes like this: OnscreenY = RealY - 0xFE * 0x100 = RealY + 0x02 * 0x100.
So the onscreen Y of all objects becomes 0x200 pixels more than it should be. And the coordinates are counted top-down.
Ok, I just thought that since the value of $CA-CB increases as you go down a vertical room it would make most sense that the value 0xFE for $CA would mean 254 screens below the top.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Great work! Taking into account the range of addresses that can be corrupted and that it's stuck to be decreased by 2, it doesn't seem like this can be used to save time. For a play-around TAS there's quite some potential though:)
AnS wrote:
Basicaly, it made the player teleport 2 screens down and thus instantly fall into a bottomless pit.
It's even more extreme than that, right? That value decreases 2 from 0, meaning that you're actually teleported 254 screens down!
Makes sense why the video Svenne posted looks like it does. He simply corrupted one or many addresses related to graphics by having the cannon shoot while falling from one of the lower platforms.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
At least got something out of a small test. You need this emulator to play it (0.98.28 to be sure).
At frame 4336, Toad's status address at $C7 changes to $7 (the death flag) but changes back to $2 (normal status) the next frame - it's still enough to kill you though. Normally when you die, $C7 stays at $7 all the way through the death animation, then is set to $2 making you restart from the beginning of the room. So the reason for why the death animation doesn't happen is that $C7 is set to $2 the frame after the death flag is set which immediately restarts the level. But why the death happens I have no idea, I'm not a hacker. If someone with more knowledge than me in this area could analyse the input file we might be getting somewhere. It's at least not related to whether you throw the enemy or not.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
werster wrote:
New Scrap Brain 2 zip, including transition to Scrap Brain 3. Not optimal since it was real time, but should be able to shave off some time with this.
http://www.youtube.com/watch?v=S7K8YF2byio
Nice one, I saw that earlier and quickly made a TAS in 13 seconds. Turned out you can do the speed trick the exact the same way on the lower wheel (that you first jump on) and break through the left wall for a quicker finish.
Nice one! I'm usually not too exited about zip discoveries like this since what it does is beating the act in less time by having less gameplay which I usually find boring. But it often turns out there are some nice twitches you can add to increase the amount of frames saved like you did here.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
MESHUGGAH wrote:
I don't understand your second sentence. You mean I do it in a way that you didn't like?
Nah, just that I'm not that interested in making TASes which require much lag management. Especially if you actually need to focus more on the lag itself than the (non-lag) gameplay. In the 1-player run, lag occured mostly due to screen scrolling, I suppose it's similar with 2 players.
MESHUGGAH wrote:
Also notice that every part of the gameplay can be switched if you remanipulate the RNG by pressing buttons on P1 (I use mostly an additional S). However L3 and L8 is an RNG madness.
Yeah, I mostly used the select button too. So player 2 doesn't change the RNG at all? If you can trace down exactly how the RNG works for Nibbly-Pibbly drops it'll surely be to your advantage (including both what color it'll have and where it'll go).
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
So then I suppose this is not the testrun but the intended final one? Judging from the lag management you're performing it doesn't sound extremelly exiting to me though so I guess it'd be ok if you did this run by yourself.
The rerecord count also suggests that it's quite more than a testrun. What counts as a rerecord in TASEdit anyway? It's of good quality too, matching my 1-player run in the areas that are comparable.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Yeah, if you want to make a submit-able run using this glitch, a playaround is probably the way to go. Here's another trick found during the making of our TAS that could be added to such a run:)
Link to video
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Svenne wrote:
Okay, disregarding the set-up, it would been without a doubt lots of time gained, with the set-up, I really don't know!^^ We haven't been doing research on setting up the actual trick, but from the start of the continue and forward! :) And, that point we use as a warp is not linked like other onces, where you do the trick and you always come to the same spot, if it is forward or backward in the game but this one is not fixed so you can decide where you wanna go!
I'd advice you to discuss this further with andrewg. He tested this particular glitch way more than me, for pretty much all areas in the game.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Svenne wrote:
So, if I were to warp from 1-1 to 1-2 and do the rest of the game, would that be all levels?
IMO, in an "all level run" the numbers above the characters heads in the ending sequence should add up to 20 - which will only be the case if you exit each level the intended way. Before such things are of interest of being discussed, though, it should be confirmed that some instance of this glitch actually saves time - I haven't seen anyone yet.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Svenne wrote:
This trick allows you to gain loads of time in a Any% or Warpless TAS! Continue was used because you wanna have the first charactor selection screen, and as Aglar said, you can do this without a continue but that means you have to go to 1-2, kill birdo, warp back to 1-1 and warp your way to 1-3 without ever stepping into that level, that was what Andrew did. I think our way is gonna make a more awesome TAS, You are done with world 1 in like 40 seconds, lol!
Actually, TASes being submitted to this site must start from power on, so the whole setup process must be a part of the movie. Also, skipping levels like this can only be used in the warped run since if you skip levels in the warpless run, it's not warpless anymore.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
This was found by andrewg, while we were making the TAS. To set this thing up, you must first (IIRC) beat Birdo in level 1-2, then go back getting a game over. That way going down that vine in 1-1 will take you to 1-3, and entering the bird head will have the same effect as if you've beaten a level with no boss at the end.
That vine in 1-1 can also take you to 1-2 if you first go to Birdo in 1-1 and then all the way back to the vine.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
MESHUGGAH wrote:
I'm a bit busy nowadays doing some MCP exams, however I would like to cooperate (with someone) mostly because I find funny manipulating the "fish breed towers" at stage 8. I'm pretty sure it could improve hundreds of frames.
I will try to do a test-run with TASedit (2 players of course) on this week after I checked that is there any difference between bizhawk and fceux latest on this game (many other games had interesting differences).
Ok, I just let you make the testrun, and then we can possibly start making a submitable run.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
feos wrote:
Aglar wrote:
I'd most of all like to see an improved version of the 2-player warpless run, since the current one have much room for improvement (especially level 8). The lag management will be very tedious tough which was mostly why I didn't do that run.
Can you please write down these improvements? I've never played that game, but it looks VERY promising in terms of strategy testing. You even mentioned you can do a quick 2P run in a few days. Also, your submission text doesn't contain any useful info on tricks.
In terms of improved strategies and tricks that I've already confirmed - there's a shortcut in the beginning of level 4 (which is used in the 1-player run), and in level 8 you can start with two extra food pieces at the beginning of pool 2,3 and 4 by eating a final one with each snake just before exiting each pool. Then I'm sure that you can gain lots of time by better precision and and minor strategy changes in the eating process.
That submission text was rather lousy, yeah. I can update the submission text later with more stuff as I quite easily keep track of things I've done in TASes, I could even add the text to a "Snake rattle 'n roll" game resource page if an editor first creates it.
I also must say that after such a long time staying away from this game, I'm sort of interesting in trying it again. The 2-player run would surely be a lot easier to make with TAS-edit than the old way I used for testing it before. It'll still most likely be a very advanced, sometimes tedious, project and very mentally challenging to do by yourself so would you be interested in a co-operation?
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
When we where TASing that part, marzojr explained it like this:
"The Target Lower Boundary (TLB) in GHZ1 is set to y=768 up until the camera goes past x=6016; after than, it is set to y=1024. When the TLB is different from the Current Lower Boundary (CLB, the one that kills you), the CLB starts to change, as follows:
TLB > CLB:
if camera Y + 8 < CLB and Sonic is on air for whatever reason, add 8 to CLB;
in all other cases, add 2 to CLB.
TLB <CLB> TLB, set CLB = TLB - (TLB mod 2) - 2;
otherwise just subtract 2 from TLB.
We are interested in the first case: TLB > CLB. If we could get Sonic on air after camera x=6016 and when camera Y + 8 < CLB , the bottom boundary would increase six additional pixels per frame for every frame Sonic is on-air. Sadly, this only happens at the end of the second tunnel. In any case, jumping when entering the second tunnel gives a 1-frame delay and sets inertia to 4096 on landing (if above it), so a slightly higher speed might work here; using this, I have been able to go through both tunnels without touching left once. My tries in replicating the trick under these conditions have been unsuccessful, though.
Relevant addresses are:
$FFF700 (2 bytes): camera X
$FFF704 (2 bytes): camera Y
$FFF726 (2 bytes): TLB
$FFF72E (2 bytes): CLB"
Shortly explained: No, you won't survive unless you slow down. You won't need to start pushing "left" until the camera position is bigger than 6016 as mentioned.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Yes it is. If the amount of frames you're ahead of the run, from which you want to hex in data, is a multiple of 32 then the animals of the later levels will sync.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Nicely done! When you say you didn't optimize the subpixels, I suppose you mean there was nothing you could've done to save time fiddling with them? If you, in the first game, have a y-subpixel value of 0-20 (when starting the jump) you're able jump up 4-block high platforms by holding the a-button for 12 frames compared to the normal 13. I'm pretty sure you tested that something similar isn't possible in this game though.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
As for getting the x-subpixels you need, I've found a way that always works - granted that it's executed right :)
If you, while falling to speed shoes, push against the left wall, then perform the glide to get into the box and then climb down to get to the left of it (which is how you did it in the gmv:s above) you'll now have an x-subpixel value of 4. After you've landed, wait 15 frames since you can't move before then. Now you want to have an x-subpixel value between 42 and 153 and this will be gotten if you for example hold "right" for either 2 or 3 frames (when I try tapping the button as gently as possible it gets hold down for 2-3 frames almost all the time, might be different for other people though). Now do a "1 rev spindash" to start falling down and hope the y-subpixels are on your side. Then do another two "1 rev spindashes" after each other. Now it's the final maneuver which can be done in one peace or in steps to make it easier.
For the risky way you need to hold "right" for 9 or 10 frames and you'll be all done, if you hold it longer than that it won't work.
If you do it in steps you must make sure that the first time you push "right" you hold it for at least 3 frames, otherwise you'll get the wrong x-subpixels. Then you can inch your way to the left side of the wall and then gently tap "right" (if you're 1 frame away from coming out of the wall you can only hold "right" for at most 2 frames, if you're 2 frames away from coming out of the wall you can hold "right" at most 3 frames, and so on) to perform the underflow glitch.
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
mike89 wrote:
So I'm attempting a real-time speedrun as Knuckles, but the Casino Night 2 level wrap still has me completely baffled.
Below I've uploaded three GMVs to Microstorage. On each of these occasions I perform the glitch to get stuck behind the box first try, each time with a full size jump. However, I can get one of three different results with the same input (at least within what can be expected from me in real time) from there and I don't understand why. Can one of the experts here look into this?
(the method once inside the wall is: spindash to break the box, fall, spindash twice, tap right, then hold right to zip)
http://dehacked.2y.net/microstorage.php/info/1658541135/CN2success.gmvhttp://dehacked.2y.net/microstorage.php/info/1588916958/CN2fail1.gmvhttp://dehacked.2y.net/microstorage.php/info/439749560/CN2fail2.gmv
Had a quick look at this and unfortunately the answer to this might not be the one your were hoping for. Which outcome, of the ones you showed, that will happen depends on your y-subpixel position.
If your y-subpixel value. when standing to the left of the speed shoes, is:
between 0-103 the result will be the one in your third gmv, you can still underflow your x-position from here though even though that depends on what x-subpixels you have.
between 104-159 the result be like in your first gmv.
between 160-255 the result will be like in your second gmv, which will always fail.
So you'll have 62,5% chance of getting the right y-subpixels in order to then give you the chance to get the right x-subpixels to break through the left side of the screen. As for getting the right x-subpixels my short test showed that if you just tap "right" gently as you stand to left of the shoes there's a high chance that you'll be getting the values for the final zip. I can give more exact data if you're still interested in trying this zip after this explanation:)
Experienced Forum User, Published Author, Expert player
(3588)
Joined: 11/9/2007
Posts: 375
Location: Varberg, Sweden
Yeah, I guess so. Earlier it has worked fine announcing it in the thread and submission text but I guess people are getting tired of Sonic runs and don't check these pages anymore :/
