TASVideos

Posted: 4/15/2011 7:47 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Too bad this hasn't progressed. But Metal Warriors' fan Nilusxyz introduced me to a new glitch, this is a big one, coauthor worthy I believe :) The trick can be seen here: http://www.youtube.com/watch?v=dvmX-jdBryc Since it was his improvement I asked him to run it for SDA, but I think I scared him with the stuff he would need to buy, although I could be wrong and it was another reason. I haven't improved it myself because I haven't had the means to record SNES for a while. The SRuns without a TAS thread made me remember about this. You basically use the spider's odd hitbox to go oob and then get pushed to the other side of walls to skip parts of the level. You jump on a specific distance so that you land slightly oob, then jump to the other "side" repeatedly (left jump, right jump, left jump, right jump etc) until you get into a position where you can get pushed to the place you want. His videos display different stuff about the game, probably you can learn something I missed.

Posted: 4/11/2011 10:06 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

jump to mugen and make ai for those characters

Posted: 1/24/2011 4:36 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

MUGG wrote:

According to gia, Yellow, Blue and Red are all the same (they use the same route in the save-corruption category) with small exceptions. Yellow has one extra tile on the map, and Red has the shortest name for the rival (making Red the 'best' version of the three). However, Red and Blue seem to lag more so that makes Yellow the 'best' version.

No :), I said they are different games, that can be completed with the same-ish strategy. Also I didn't say yellow was the best version, just the one I did the tas with, (I like the pretty colors). Hmm, unless I didn't type that and you are correct in the end, but that's what I recall. I did something similar, slower because I went warpless: http://www.youtube.com/watch?v=ts8Axz7VrfE, I doubt something like this would be accepted by you though, you got several categories already. I haven't tried to do the run on other games but yellow, so I can't tell if they are optimal if you view them all as a single game. I don't hence I didn't. The ram offset could be useful, or not. You are counting on p4wn3r trying every variation, even if he confirms so you should only be sure after you confirm it yourself/got proof :D

Posted: 12/29/2010 6:55 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

DarkKobold wrote:

sgrunt wrote:
notably, getting to the hall of fame seems to use a trick fairly similar in nature to what's ultimately used in the Yellow run.
I couldn't help but feel the same way. I basically skipped around to the end, only to see the exact same glitch used in Pokemon yellow. Yes, it does it without reset corruption, but that appears to not actually change the glitch itself (changing exit location by discarding an object), simply the pre-cursor steps to executing the glitch. While a good set of glitches, I think the warp-to-hall-of-fame glitch shouldn't have been used.

Different glitch actually, in yellow, the "glitch" is corrupting the save. From there there aren't glitches, just a "hacker" exploiting the hole opened. Here the glitch is the ZZAZZ glitch, to the viewer it may look the same but it isn't. Just looking at the RAM makes this clear, the effects they have are very different. Although once the hole is open you do pretty similar stuff with it. But again, you dont do save glitch here, you dont do ZZAZZ there.

Posted: 12/28/2010 11:33 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

MESHUGGAH, I think I get your definition but I am not sure if I really did. You'd be fine with the statistics and /or credits looping infinitely since that's an end game state and you cant progress further if it loops infinitely. I'm not sure if you'd be fine if the game was forced to loop infinitely drawing the last frame that says The End. You are at the very last instruction of the ending, and you are constantly executing it again and again, does that count as state? You haven't just drawn the end once and continued to play afterwards, the game is locked drawing it forever. However if you created the code that will make this loop in some area of memory you have access to, and as such it is not the original's game ending code the one being executed, it is not the The end that should be drawn, its an identical clone you created, I'm not sure if it would fit your definition. Warp, assuming such game can be defined an ending, reaching that final stage would probably be it. Assuming again this final stage is open ended, as in no sequential rooms, you can go up or down but everything's empty, once you reach the stage you can't be sure where to proceed. Personman, I suppose that's an option, I recall a tetris tas that set maximum score as it's own ending goal.

Posted: 12/20/2010 6:36 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

I kindly asked for a frame count before submitting. Question, when you reach the hall of fame then after talking to oak and the screen changing you are given oak's final rating on a special sequence, after this rating the credits start. If you dont reach the hall of fame with "152" pokemon then the rating will most likely have two bodies of text, which means you have to press a button once to continue. Am I correct thinking this button press would be part of the tas (and count toward frame count) or has the ending started before this part? Are there other movies where the viewer has to press a button to complete the ending sequence?

Posted: 12/13/2010 4:57 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Patashu wrote:

Do you have any games in mind which are ambiguous?

Pokemon, it's on the post. RGBY, but Yellow is more specific if you like.

Kirkq wrote:

I think having some viewer sensory confirmation that the ending has been reached is a necessary requirement. This topic could be a ridiculous debate on the workbench for sure.

Yes, if you descipher my monologue I wonder whether sensory confirmation is a must or completely irrelevant. But then I worry, on either case, how much confirmation is necessary.

pirate_sephiroth wrote:

To take the game as beaten, I think we need to see the expected conclusion. Just making the game logic consider it beaten is not enough and probably not even visible in a video recording.

Ok I'll take it you prefer sensory confirmation as well, but what is exactly the expected result? Yellow's ending has several parts that have some visual or aural feedback to the player. Perhaps just rolling the credits is enough for you? Would it be enough with just the last frame of the credits? Do you require the whole thing since Oak's speech? what if he doesn't give his ending speech, and gives his initial speech instead?

p4wn3r wrote:

I made sure to put "Aims for fastest completion of the game"

Maybe I understood you wrong, but first you said the runs have to be entertaining. Now this is another debate, and my opinion may not be the common one, but if you go for fastest completion then you will not compromise even once and will sacrifice entertainment wherever necessary. The problem here is when you compromise too much and overshoot and end up not actually beating the game. As for the rest of your post I see you also decided to attack just setting flags, so I'll file it under "an ending requires sensory confirmation". For all we know many of the extreme possibilities could be slower from a tas perspective and as such much debate saved, but I still would like some opinions.

Posted: 12/13/2010 7:35 AM

Post subject: How to define a game's ending?

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Hi, on average gaming defining "beating the game" is easy, but lately I've been struggling trying to figure out when I have crossed the line and have not actually reached the ending. Haven't found any discussion about it on the internet, although it probably was a bad search query... You know pokemon, and how you can manipulate ram directly, so imagine you can do absolutely anything you want. You also know how it plays, its classic for many games: You play through several maps until you reach a final map, usually have a boss fight on that map (in pokemon you have the fight on the previous map), play a "cutscene", in this specific case there's a white out and a special rating screen (in other games it could be a statistics or score screen), progress is saved, setting flags for "post-game" play (a quest flag, access to Hall of Fame on the computer, etc), then the credits animation rolls in, then the game resets cleanly. All of this is part of beating the game for the average player. However we can skip the part where we play through several maps, and just reach the final one. And its generally accepted that we beat the game. We can skip the part where we fight the final boss, I think most people would still accept that as completing the game. We can skip the part where we enter the final map, at least visually. Some might argue if we don't reach the final map at all, but we are still doing the steps that follow it. We can skip the cut scene, the statistics screen, game being saved, flags being set. I personally would end up accepting it just with the credits animation rolling and the game resetting. But what if to all the previous, we also garble up the graphics of the credits animation in a moderate way, we see some developer's names but they are barely visible in a corner. Let's pretend this is still valid, so now let's completely garble up the graphics, we see stuff moving with the speed and direction credits would roll, but can't tell for sure it's the credits what we are seeing, we still accept it's the credits because the bgm is the same as the one played then. Remove the bgm. Still acceptable? If so, remove the part where the game resets, have these so called credits loop forever. I could barely get up to this point, but we can still make it more extreme. Instead of playing credits, have the ending flags be set silently, the game goes on but you now have access to post-game material. Did we suddenly beat the game? Or instead of a loop have the credits end and the previous game continue like if nothing had happened. Would this be beating the game or did we just play an animation like it seems to have happened? But this is THE credits animation, sure it doesn't look like it, but data was accessed, data that is only accessed when you beat the game. But then comes the problem, how much of this "special" data has to be accessed for it to be considered beating the game? what if we only accessed the last frame of the credits animation? What if we only accessed the last couple particle effects that are created on this animation and nothing more? What if we see absolutely no indication that we beat the game, however we can prove that we accessed at least one instruction of the procedure that would be called "BeatTheGame()" if we reverse engineered the game's code, hell what i this specific instruction is its final call, the one that resets the game. What if it doesn't reset the game, and it looks like the game crashed, black screen, no sound, no response? What do you think?

Posted: 9/8/2010 10:26 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

what do you mean, something like C:\> vba.exe -playmovie '<file>' ?? and then it opens the emulator in window mode and loads the movie? If so why dont you just edit the emulators so they have that capability, or is that what you mean by hack :P

Posted: 6/9/2010 1:28 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Get z80080.pdf: Zilog Z80 Family CPU User Manual User Manual 308 pages It has a reference of all the Z80 instructions, using the debugger it tells you the instructions that are being executed, so then you can figure out what is being done. What emulator/debugger are you using? I used bgb I think. Is Rhydon species 00? I thought it was 01, 00 is the first index on a list. In any case maybe you are right and it is a trainer picture.

Posted: 6/8/2010 7:19 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

lol right, eh, that player is for a "new" (not really) game engine ala game maker, torque, etc The engine and tools are pretty cool, the developer makes a game and they can package the same project as a desktop application, web (like this one), iPhone, even the Wii. It's fine to rather not install the web player if you don't expect to hunt games that use this engine and that also happen to use the web publish option for packaging, I already had it installed from their engine trial.

Posted: 6/5/2010 5:12 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

never played zelda ii but managed to progress a bit, lost to the sword armor

Posted: 6/5/2010 5:08 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

last place you visit with an object that disappears, before you initiate battle, will have that object disappear, apply to question: route 24 or cerulean's, you can always try and figure out for real

Posted: 5/21/2010 7:19 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Yes sort of, I don't know the common algorithms for video/image compression besides the simple differential compression or whatever its name is. But if you read HyperHacker's comment he makes it very obvious a very likely reason, but you will have to find the "image" and also know how to decompress it. Let's assume he is correct and let's make an assumption on how the decoder could work. The image lets say starts with 20 followed straight by the decoder "instructions", no body length for some reason, after all of these are read theres a 50 to signify EOF or end of image. ZZAZZ works with the last "pokemon" indexes, like pokemon 251, which means there's no valid pokemon after it. So its very likely there is no EOF around there. Now lets assume the decoder works like this, read first byte of first instruction, this is the instruction, then process the instruction. When done read next instruction and so on. Lets say the first byte is 55, then 55 is the instruction and to the decoder it means read the next two bytes and jump that number of bytes in memory, so it does, and somehow it matches the area of the overwriting ZZAZZ effect. Then the next instruction starts with 56 which means read the next three bytes, the first will be the number to write lets say 99, the next two the number of times to do this. Then it goes and writes 99,99,skip one, that number of times. And you end with the ZZAZZ effect. Then at some point it reaches an EOF before crashing the game. Of course this is incorrect and you won't be able to know for sure until you figure out how the decoder works and where is the image. If it wasn't the image it could be the pokemon's name, remember ZZAZZ isn't missingno., it is a glitch pokemon at index 251 or more. When reading text there are 256 possible values yet not that many letters, some values are control codes for something else, like when sounds are played or the volume modified, or whatever, so the ZZAZZ effect could be triggered by "printing" these invalid names. Perhaps you can test this easily, copy the name data for pokemon 250, 251, 252 etc until you find the next "50" (and if it is close then just keep copying until you think its enough) and put all that over Magikarp's name or whatever then find one in the wild.

Posted: 5/20/2010 9:53 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Er, are we casually chatting or are you really not understanding and asking something? You talked about memory map, that's memory. You want to find the "image" in the ROM, not memory (not entirely true, but to get started yes, find them in the ROM first). Since all the images wouldn't be loaded to memory just the ones on the required chunk and only during the "load image function" then the chunk would be unloaded and a better chunk (like battle gui code) would remain in memory most of the time. So just load the ROM on an hex editor and treat it like the memory viewer, you can even load changes on the emulator and see what they do :P Yes, Missingno is a placeholder, but not in a separate routine that does "if pokemon not found then go to this specific placeholder", no, the routines just take an index (species number) and search in the data as if it was an array (start + offset*index), and there happens to be values there and they get loaded, the programmers did write MissingNo. in many of them (in some didn't it seems, ending in glitch names), but they didn't give it much thought as a pokemon it seems because all they created for it is the name, there's no sprite, no moveset that makes sense at least, anything that I know. On later gens they created an sprite at least :P No idea about what the "something else" is, it's possible it's nothing. They could have had all the pokemon data area cover the 256 possible pokemon and just left the data "empty" on the unused spots, and by empty I mean they didn't touch it anymore after development finished so missingno moveset could be leftovers from development. OE 81 is not ascii for B, it is B in pokemon terms already, I believe the ROM should have the string. Yup it does, all the pokemon names are stored on an array, so I assume the base stats have their own array and all the stuff separately instead of an array of "struct PokemonSpecie". It starts somewhere at 1BFFC and I could find BULBASAUR and later two MISSINGNO. right next to each other followed by GROWLITHE, it seems they have 50 to signify end of string but if the string is too long they just cut at "max length" of 10 and omit the 50. That comment by HyperHacker is interesting, but I dont know how the decompressor works to help you with that. If you can't find anything even among rom hackers maybe just maybe loadingNOW. is still active, I think I last saw his nick here http://www.filb.de , IIRC he used to obtain the images from pokemon games, but not sure if he tackled RBY.

Posted: 5/20/2010 7:44 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

for new speedruns, read the news at the main page

Posted: 5/19/2010 8:09 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

It obviously does because the area of memory belongs to something else and the game just reads it as an image. I recall a program that could extract the images from gsc it showed the decompression algorithm, could be out there if you search. The names are most likely stored as they are used, i doubt they are compressed, I am not sure though, If you can't find bulbasaur try finding its base stats. I got an incomplete memory map, which is different than a ROM map. To find images and stuff you'd want to check the ROM instead of the RAM. The ROM is bigger than the GB RAM, so the game isn't completely loaded into RAM, instead pages of it are flipped (I dont recall the size of each page) into memory when they are needed, so you'd need to use a debugger and breakpoint emulator to catch it just as it loads the page with the pokemon names and image data. Or like I said just use an hex viewer on the ROM file. Glitchcity has a ROM map iirc also datacrystal and possibly rom hack sites and/or their authors.

Posted: 5/3/2010 12:11 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Having never played the game to me this is a platformer not a racing game. I got bored on the fifth world, but the last two were great. I only didnt understand why you slowed down on one of the worlds and didnt jump but comments explain it.

Posted: 5/2/2010 2:21 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Graveworm wrote:

I've only played like old SNES games the last few years, but I guess it's hard to make the map look like a globe with that kind of technology, hence you'll have to imagine the world on the map being round and all edges connecting in some kind of twisted way instead. Like papier-maché on a balloon. (And extensive cutting.) I probably missed the point by a mile though. (The center of the map is one pole and the edges are another, do I win?)

no, that would be a solution, although if you mapped the earth like that it would look weird, and so would a compass :P

Posted: 5/1/2010 10:35 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Never thought about that lol.

arflech wrote:

I first got suckered in by the "Fake Difficulty" article in the description for one of the runs. Anyway you could make the maps more realistic by having all users who walk up the top of the map end up walking down from the top of the map, halfway around it, like this: img

Or just have a separate area for the north/south pole so that you go there when you reach the top/bottom of the world map. Then you can choose where you spawn back depending on what direction you leave this area. Or use two world maps, one each for a half of the world.

Posted: 4/20/2010 4:26 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

route 24 has that pokeball that can disappear, and you go one step into route 24, so ill guess that pokeball is the one that disappears I repeat, for saffron use vermillion, there are trainers you can fly away from on the routes next to it. if you want nothing valuable to disappear yes cerulean should do since you probably made it disappear already while playing

Posted: 4/19/2010 10:01 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

in cerulean city there is a tree south and also the rocket that gives you dig and the police officer that moves, gary and stuff, not sure which count or not but im almost sure cerulean does have one of those because I dont recall snorlax working when I used cerulean's pokecenter vermillion was safe though

Posted: 4/17/2010 5:26 PM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

I enjoyed it, description should describe what each minigame is supposed to represent, there are a couple I didn't get. Or a link to a FAQ, that would help.

Posted: 4/12/2010 12:13 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

Ok for the modified TAS Movie Editor, just go to its own thread and download from sourceforge since those changes are on the main program now. For the vba bot that plays the generated movies use this: http://rapidshare.com/files/374817440/vba_bot.rar.html If you need the gens bot let me know. For all these programs open the readme/config/etc files with notepad to check the configurations and the comments on how to use them, I think I made them detailed because I knew I would forget how they work. You may also read the King's Bounty (genesis) thread. After you do those if you still don't understand something send me a PM.

Posted: 4/7/2010 1:14 AM

Quote

gia

Experienced Forum User, Published Author, Player (109)

Joined: 5/3/2006
Posts: 223

That's too much I think, when it happened to me I just took note of the achs so I can "get" them later without the unlocked message. I managed to get mozilla's javascript engine working this time on snes9x, and also can contact my server and login. So technically everything can be done now provided I don't introduce memory leaks with the js engine. I need a way to match a rom loaded to a unique identifier that represents that rom (included its unique memory map), since a script would not work if due to version differences or ips file the memory is mapped differently. Any ideas? CRC is unique most of the time, but what if it isn't... Or do I go simple and just use the filename? If someone can make a javascript code/pseudocode with memory addresses to check for something simple that can be "achieved" multiple times with ease on mario or some snes game that would be cool.

Posts for gia