Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
TheMG2 wrote:
You misunderstand me. I do not doubt that it has wow factor on it's own. I don't dislike such glitches in general. I just don't think that low% would be the best way to show such a glitch off.
low% is quite possibly the only category in which missile overflow can be showed off, as it is questionable if it saves time in any%.
TheMG2 wrote:
It seems pretty clear that they would much rather do a TAS without (or two TASes, but noone wants just the TAS with overflow), especially if they think a TAS with would not be received well.
Uh, that's not true. I personally would much rather see a TAS with overflow than a TAS with a very prolonged Ridley fight
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
TheMG2 wrote:
Having a ton of ammo goes against why many viewers would watch low% in the first place. A lot of people watch low% with the expectation of the use of less equipment, not the use of more ammo that is traditionally obtained through item collection. If I were a casual viewer, I'd feel a little cheated by seeing such an underflow glitch used in a low%.
I have no comment about using it in other categories however.
Idk about you, but w/o knowing about missile overflow my first reaction would be "wtf how does Samus have so many missiles!?" rather than "lame, no true low% run has thousands of missiles." The awe factor is more important to the average viewer than the meticulous routing, and pumping Ridley and MB full of super missiles certainly has awe. And if we absolutely need a run in which Samus gets by by the skin of her teeth, then we already have RBO anyway.
Either way, it's neither here nor there. What is entertaining and what is not will be judged when such a run is submitted, and if missile overflow does not result in an entertaining run it will not be published.
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
boct1584 wrote:
Also, I'm fairly sure it should be called "underflow," as the numbers go below 0 and jump to a very high number, similar to horizontal zipping off the left edge of the screen and appearing on the far right of the level in the Sonic games, which has always been called "horizontal underflow." "Overflow" would be when the number goes too high and drops to 0 or goes negative, such as when you kill the Lavos Core with an Elixir as in [1285] SNES Chrono Trigger "save glitch" by inichi in 21:23.98.
It shouldn't. Underflow means arithmetic overflow, which describes a phenomenon with floating-point values
Integer overflow is both ways. Think of it as carrying from a bit outside the number. I.e. when subtracting 1 from 0 in a 16-bit int, the 0 is treated as 65536, or 0x10000. So it makes sense that you would get 65535 (0xFFFF), the maximum value a 16-bit int can hold, out of the operation
---
On-topic, though, I don't see how missile overflow can be considered a major glitch. It can be considered memory corruption if you really stretch the definition, but it doesn't corrupt arbitrary addresses; it only corrupts one. And it doesn't skip a huge part of the game, either. In fact it skips nothing, since you still have to fight Ridley and Mother Brain. So it really boils down to just entertainment, and if missile overflow is not entertaining then it won't be accepted because the Vault doesn't take low%.
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
I wrote a program to help with Core-X fights. It takes initial position and target position as inputs, and outputs all potential solutions to get there on the right frame. See --help for details
https://github.com/Anty-Lemon/CoreXFucker
According to my testing, the ideal x/y positions are ±5 from the position of the target enemy near the center of the boss arena.
If you want to find the ideal positions manually, they can be found by cheating Core-X ptr + 0x2E-0x31 to 0, then poking the defeated Core-X's position until it stops flashing
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
BioSpark wrote:
Anty-Lemon wrote:
While pauses are used throughout the run to manipulate the RNG, they are not used in various other places where it would save one in-game frame (the second room, for instance).
Pauses can never save in-game time in this manner. Depending on the parity, you'll either lose 0-1 frames or 1-2 frames.
That's not right. To elaborate, doors have a frame rule which is dependent on the game's global timer at 0x3000002, not the in-game timer. If the frame rule is not in your favor, you can pause for an odd number of frames to make it work. Try it for yourself
EDIT: Ohhhh I see what you're saying. I forgot pauses waste an in-game frame
BioSpark wrote:
The next 0% and 100% runs must aim for real time anyway, since there's a memory corruption glitch that allows for a very low in-game time. Both runs will be over a minute faster because of new tricks (and new route changes for 100%).
Yeah, at the time I made that post the memory corruption stuff hadn't taken off yet
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
Temp encode for those who don't like Twitch and/or floating heads:
Link to video
Voting yes 'cause I (surprisingly) enjoyed it. The 100% overlap is there, but it was still entertaining imo. Tho I admit I haven't watched 100% in a while
Plus, it was nice to see the Etecoons (whoops, already in RBO) and Dachora for once, even if you don't save them :P
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
smellyfeetyouhave wrote:
I've been playing around with Hourglass a bit lately and after trying it in both Virtualbox and VMWare, I've come to the conclusion that VMWare's DirectX support must be significantly better than Virtualbox's as I've had a LOT less random issues with Hourglass in VMware over Virtualbox. Has anyone else had this experience as well?
Yoooo! TASing the Undertale demo?
I tried it in VBox, and the only issues I ran into were severe slowdowns. I'm not sure if it's related or not, tho
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
MUGG wrote:
I'm currently trying to convert this vba lua script to bizhawk.
http://pastebin.com/mgSCLgwJ
But when I run drawRNG(), it prints a ton of out of bounds memory read messages and it basicly frozen. It displays something every 10 sec though.
Help?
You forgot to set the memory domain to ROM in actorenemydata()
Also, ENEMY_DATA_START should be 0x500a98
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
Gamer Maiden Sonia wrote:
I wish this emulator could at least .avi dump, but it can't even do that. So it's useless for TASing.
There's no way to record gameplay footage unless if you use an external program like FRAPS or something.
While it really should have avi dumping, kkapture is a perfectly fine alternative
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
BioSpark wrote:
I'm not sure what you're trying to say, but there isn't any issue with manipulating the eyedoor.
Eyedoors have a timer that starts with 60 to 120 frames. If you're not in range when the timer hits 0, the timer resets to another value between 60-120. The goal is to get in range of the eyedoor right before the timer runs out. Manipulating eyedoors involves waiting before entering the room to get a favorable set of timer values.
Here's a script for eyedoors, if anyone's interested:
http://pastebin.com/DAGnjAdJ
I know all this. I've also written my own script for bizhawk :U
I was saying that the only time there would be a problem with the eyedoor is if you could get to it before 60 frames had passed, and in this case it takes about 4 seconds
EDIT: Also, your script is slightly inaccurate. The eyedoor uses (samus.y + (samus.topOffset + (samus.topOffset >> 0x1F)) / 2) as samus's y location rather than simply samus.y. The script is also a bit naive by assuming that being behind the eyedoor can also result in Samus being in range, though that shouldn't ever matter
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
jlun2 wrote:
I'm not sure if it's answered before, but for these functions
event.onmemoryexecute
event.onmemoryread
event.onmemorywrite
How exactly do I know which one to use for an address?
Usually you shouldn't use any of them. These functions are for disassembly, and/or games you've already disassembled. Just reading the address every frame should be fine for most purposes.
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
ISSOtm wrote:
According to TheZZAZZGlitch's video, when creating the save, the first segment is chosen randomly, and each save shifts every segment right by two segments (ie 23456789ABCD would turn into 0123456789ABCD).
For normal players, there is a 1/2 chance that 0 will never be the first segment no matter what. But no video game randomness is TAS-proof, so that's no problem for us.
What? It isn't a valid save file if any sections are missing. It would be 23456789ABCD01.
Anyway, I cannot reproduce this. From my testing, every two saves always shift the sections right by two, and the second save always starts on 0xC. My initial testing was on no$gba, but I tried it a few times on mGBA to confirm.
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
grassini wrote:
the fact that they have a model for munchlax on XD doesn't mean it has an adress as usable poké in a 3rd gen game,it's like saying any actor model in the game necessarily has a pokémon counterpart in the adress table,if that's what you're implying.
Sorry if this goes offtopic,but back in topic,is there any expectation that these tricks might be usable in RTA runs or is it TAS only because of the memory allocation changing for the saved files?
Idk if this is accurate for console, but on emulator the save always starts out with a certain layout. I.e. the first save starts with section 0xD, the second starts with 0xC, the third with 0xB, and so on. After the first section, the sections ascend, looping back to 0 after 0xD, until all sections are written.
Therefore, if you save the same number of times every run, the savedata will always have the same layout, and thus ACE will always be first try.
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
Metarkrai wrote:
The 0x65C you want, is it a Move ID or a value you'd like to form using a nickname ?
Would your method also use Glitch Move with animation pointers on saved data, or would you use a pointer to the nickname of a PC Pokémon (in RAM) ?
Decamark, using the old box method. It's the same decamark ZZAZZ used in his original ACE save file. I'm just having trouble glitzer popping it :/
It could probably be repurposed for use with a glitch move, but I wrote that code before this was discovered
Metarkrai wrote:
Would it be possible to program longer jumps (mainly towards PC Items) using nicknames, or would the amount of characters limit the jumps too much ?
Depends on where you start from. Using relative branch instructions, you can go pretty far, but like you said it's very restricted
Another possibility is jumping to box nicknames instead. The way box nicknames laid out in the savedata, you could start from box 1's and end at box 14's without ever jumping in-between. Plus, you don't need to catch any extra Pokémon
EDIT: Oh, or you could just start in save section 0xD and execute a ton of 00000000's until you reach the box nicknames
Metarkrai wrote:
If you could use all possible values for every byte of your instructions (instead of only values from characters), would that reduce the amount of required nicknames for such a corruption ? (or for a corruption that would set a specific flag, or change a specific byte)
0x1
It could potentially be pushed lower, but the only improvements I know of are converting it to TAS strats. I haven't explored the possibilities, but at the least LDR and SBC could be replaced by a single MOV, provided you know the new checksum. Also, SWI could be replaced with a hard reset, bringing it down to 5 instructions
Metarkrai wrote:
Also, as credits seem to be triggered by the command special 0x113 (accordig to XSE), would it be easier to directly trigger that command/thing instead of performing a teleport towards HoF ?
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
Metarkrai wrote:
Oh yeah, why not. I didn't think about it.
The required code would be shorter and could maybe be made with 1-2 Pokémon only.
Not necessarily. HoF warp is doable with much less code than what ZZAZZ has shown. I've got code that should work in RTA (if only I could get a damn 0x65C) in 7 nicknames by reading the PC instead of using a bunch of clumsy adc instructions. All it requires is that the 5th party slot is empty
0E003300 b #0x2FFAC ; ****uE v
0E0332AC movs r11,#0xB10 ; “D…ouE v
0E06325C adcs r10,r15,#0x3B80 ; z.EnuE v
0E09320C ldr r12,[r10,#0xE0]! ; lF/quE v
0E0C31BC str r11,[r10]! ; …9quE v
0E0F316C sbc r11,r12,#0xB10 ; “DRnuE v
0E12311C str r11,[r10,#0x3C0]! ; F‘9q
0E123120 swigt #0xDFDF ; kk U
Encoder, Experienced Forum User, Published Author, Player
(35)
Joined: 10/22/2014
Posts: 425
MUGG wrote:
Can't you just emulate XP in virtualbox and run hourglass in it? That's what I tried, but my games of interest don't seem to work with hourglass.
In my experience that is waaaaaaaay too slow
Warepire wrote:
They do work on Win7 and Win8 (not sure about Win10), but they can crash the computer (not the game, not hourglass, the computer), the frequency depends on the game.
I have never had savestates work on 7. I guess my computer just hates Hourglass, or Hourglass hates the games I've tried
