Posts for mkdasher

1 2
6 7 8 9 10
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Stevmay09 wrote:
[quote="mkdasher"That's pretty good! I give props to you because i tried to make a TAS of this game as well. I stopped because i couldn't figure out how to manipulate seeds to get the pokemon I wanted. (I'm also not very familiar with scripts). I can't wait to see the improved version! How much time do you think you could save?
Not really sure how much I can save. More than 20 secs at least, but I don't know. Most of the improvements I can do are very small. Of course finding a new route or something could increase that number.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Stevmay09 wrote:
Wow! Can you post the file so i can see it? What was the in-game time at?
In-game time (according to hall of fame debut) is 1:43:54. I've been testing with seeds, and trying to figure out how to get that seed that would give me only 2 encounters until Pewter City. I found that the seed I want is 0000A97D. This way I get 4AD331AF on the first RNG address, and 00004AD3 on 2nd RNG address (00004AD3 is the value i want to get just 2 encounters). I tried bruteforce searching, but yeah, internal loadstating and savestating with lua doesn't work properly. (I tried it, found the seed, but then desynced when replaying the movie). I won't be able to figure out how that seed is being generated, so my only idea was doing a bruteforce search. So I would like to know if there's any other way to do it. If there isn't, I guess I'll stick with 3-4 wild encounters in the end, but I want to keep trying this. EDIT: vbm file of the test run: http://dehacked.2y.net/microstorage.php/info/532642281/PokemonFireRed%20-%20mkdasher.vbm Emulator: VBA-RR v23.6 svn461 ROM: 1636 - Pokemon - Fire Red Version (U) Real time clock enabled, and Flash 128k. As I said, the test run has obvious mistakes, such as buying x sp atk and not using them (I didn't know I wouldnt need them until I finish the run).
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
I just finished the any% run. Final time was 1:47:46, 15405 frames slower than japanese version. (japanese version saves a lot due to text). I lost some frames in the Elite 4 because I was doing it very quickly and this is a test run anyway. I used some strategies that are surely faster than the ones used in the japanese version, for example using rattata in stead of digglett for dig and cut, saving a few critical hits due to better IV, and more. Now I'd like to start again to see how much I can improve it. I know a few improvements that I can do. These are some of them: - I want only 2 encounters until Pewter City, and those 2 encounters must happen on Route 1 (second time), or the second one on Route 2. I found there's only one seed that would give you that possibility. However since I have no idea how to manipulate the only chance I have to find it is doing some bruteforce search. I did a search test this morning and after 10000 seeds I couldn't find the seed I want. There's 1/65536 chances of finding that seed so I don't know if I'll be lucky with this. If I can't do it, I'll try getting 3 wild encounters until Pewter City (current test run has 4 w.e.) - I don't need x sp.atk, torrent can do the job. I can lower my HP after battling Gary at Silph C.O. and mantain my HP low enough until the end. Just have to make sure I have enough HP for Gary's Arcanine (last battle) since it'll use extremespeed, but that's not hard. - After doing the test run, I know how many items i'll need during the run. Buying 11 items in some cases will save time to buy as soon as possible. In other cases it'll save time to just buy enough items, saving frames in item menu selection.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
FractalFusion wrote:
mkdasher, I watched your run. Since you're doing the Round 2 route, when do you plan to catch 60 Pokemon?
At first I thought about doing Round 2 but then I thought that it'd be easier to start with just an any% run, since it's the first time i'm trying to TAS pokemon, so this is an any% route. And yeah, the way to survive Raichu is to have more health, that's why I'm using that second potion. EDIT: I found out that the best RNG seed for wild encounters is 00004AD3. I still have no idea how to get that though. I may try to do some bruteforce searching when I'm back from holidays. This seed will give you exactly 2 encounters before Pewter City, and those 2 encounters can be done the second time you go on Route 1. EDIT 2: Sorry I said it wrong. 00004AD3 is not the seed, that value is the first RNG value that I want.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
FractalFusion wrote:
mkdasher, your WIP desynced on me. Which emulator and ROM did you use? Did you have real time clock enabled? Did you use Flash 128K?
Emulator: VBA-RR v23.6 svn461 ROM: 1636 - Pokemon - Fire Red Version (U) Real time clock enabled, and Flash 128k.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
I'm posting a WIP, since I don't know how long it will take me to finish it (if I am able to finish) http://www.mediafire.com/download/t2a67de6mu1ira1/PokemonFireRed_-_WIP.vbm (Didn't upload to dehacked microstorage because it didn't seem to load for me) I used FractalFusion's lua script most of the time. Well, I actually extended it to see when to hit Max Damage Non Critical Attacks (I need some of those in the run), Side Effects, and for Abra. I do not know how to luck manipulate the seed created on Charizard screen (wild encounter RNG seed). Still, the best thing I could do was testing a bunch of seeds I was getting, and then I made a simple C++ program to check which seed would give me better results from Pallet Town to Pewter City. The best I could get was 4 wild encounters (capturing 2 of those), although I'm pretty sure it must exist a better seed. Squirtle IVs: HP: 29 / Atk: 31 / Def: 4 / Sp.Atk: 31 / Sp.Def: 9 / Speed: 29 Nature: Lonely (+atk / -def) About the run: - This is a WIP of a any% run. So I will just fight the Elite Four once. - The WIP is done just after clearing Safari Zone. - Rattata is used for Dig and Cut. Using Digglett is slower. - 2 potions are used. First one is to survive Misty, 2nd one is to survive Surge's Raichu (it'd spam Shock Waves with lower HP). - I get advantage of Torrent Ability when I can. Some extra things: - When entering on a new area, the wild encounter's RNG will not advance for the first 6 steps. However that sometimes don't happen (as you can see I was able to encounter Abra after 4 steps). So what I think is that for the first 6 steps, you can get wild enouncter's RNG advances, and I'm pretty sure that's determined by the other RNG address. Also, there are areas (such as Pokemon Tower) where the first RNG advance can be delayed for more than 6 frames. - RNG seems to be different after the first time I enter in a Pokemon Center. Before that, RNG does 2 cycles per frame on a battle and 1 cycle when not battling, and after entering for the first time, RNG does 3 cycles when battling, and 2 cycles when not battling. The second one is easier to luck manipulate.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
I think I found a walking / running timesaver (I think this has never been done before): Link to video It's just walking to the direction of the person / wall for one frame before changing your walking /running direction. Saves aroun 8, 9 frames i think (havent checked exactly how many). The reason of this is that it avoids that slow first step that the character makes to turn.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
I got back to this. I have 2 badges, and going to Bill's house. I think I'm going to have a problem with PPs but I dont know how to solve it yet. I'm going a bit fast because this is the first run I'm doing and I want it as a reference in case I decide to do a good optimized run after this. Still, I'm optimizing the best I can, just not fully sure about the route. Edit: Reaching Celadon City with no PPs left, I guess I'll cure my pokemon in the poke center there. Tested the difference between teaching Bite or Mega Punch to Wartortle. Mega punch resulted to be slightly faster.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
MrGrunz wrote:
sonicpacker wrote:
Pheenoh wrote:
MrGrunz wrote:
Is there some secret sub 5 TAS in the making, that noone is allowed to know about until it is submitted, so you can blow all our heads completely away? Yeah, that has to be the reason why nobody is answering my question, I'm sure.
Or maybe those working on it are too lazy to post?
Or maybe...no one is working on it. <_<
but still...is snark's most recent run still improveable using the bowser dance and maybe some other stuff?
snark's run is improvable using bowser dance for 1 frame assuming it's possible to luck manipulate bowser without losing any frame before that, which hasn't been proven yet. There's a very small margin on luck manipulation before that.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
FractalFusion wrote:
For seeding purposes the value at address 0x4000104 is used. Here's the bad part. That address is opcode count. That means that potentially any action can influence that address. Furthermore, I believe that even the smallest core differences in emulator version will result in different values. Unfortunately, internal Lua savestating and loadstating simply does not work (as intended at least) in VBA, although there are workarounds.
Yeah, I knew it was because of that 0x4000104 but I couldn't understand how it worked, thanks for explaining.
FractalFusion wrote:
By the way, the game determines which wild Pokemon appears by calling an RNG value (mod 100) and finding the slot of the table in which it belongs. Each area has a different table. Rares are mostly valued 98 and/or 99. You can hack my lua script and change one of the tables so that it colors whenever RNG value is, say, 99 (mod 100).
Thanks, that's what I needed to know. I'll try to add that to the lua script when I have to try catching rare pokemon.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Link to video Did a little test run. I'm doing the Round 2 route, so I have to catch 60 pokemon. I just don't understand how seeds are created (The one after charizard screen and the one after name screen) so I just test a bunch of them and then simulated it to know which results they would give me. Since it's a test run I didn't pay more attention but for a good TAS I'd need to know how to calculate all the possible seeds in an X number of frames. Here I got 4 wild encounters, I avoided getting a w.e before having pokeballs. FractalFusion said 3 was possible, and I actually think 2 is possible as well (at least in theory), but I just don't know how to get a better seed, I'd need to know something else rather than just testing. Squirtle IVs: 19 HP / 30 atk / 25 def / 30 spatk / 20 sp.def / 30 speed Lonely nature (+atk/-def) Not sure if a +sp.atk nature would be better, I still have to test. I used FractalFusion's script for critical hits and stuff, I still don't know how to manipulate secondary effects like lowering speed with bubble. (Just doing it by trial and error at the moment). Also I still don't know how to manipulate which pokemon will appear on a wild encounter. I guess getting pokerus would be nice too, I just havent tested it yet. And just to end with, I understand how MID text is better for luck manipulation than FAST text, however MID text loses one frame each time you start scrolling a text, so I'm not sure if in the end it's much better. Well, at least is better looking I guess..
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
I recently decided to try learning some Pokemon TASing so I decided to try stuff on this game. I've been testing FractalFusion's script (really good one btw, you did a really good job with that), and trying to understand the RNGs. There's a thing I don't really understand at the moment. When RNG2 (encounter RNG) is seeded by RNG1 (just after Charizard screen), RNG1 doesn't change like the formula says, and I guess that happens in some other places. Is it known how to calculate that changes? I couldn't find information about it. Edit: my bad, I read it wrong, I think I understood what it makes it change now, but still I don't know how to manipulate it correctly. Well, I'll try to learn a bit more, hehe.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
This is the BitFS improvement for 0 stars: Link to video Kyman got the idea of the improvement, which consists on dive grind after the pole glitch, in stead of wallkicking. It saves 4 frames. Then I think those wallkicks are very good optimized.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Yeah, BLJ strat for Big House in the Sky is still faster for a 120 star run.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
I'll watch this as soon as I find a free hour.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Congrats Nahoc! well deserved man :)
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Voted meh. Hope this gets improved a bit, I know you can do it ;) But well still nice to see this game TASed.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
MrGrunz wrote:
Suggested Screenshot (I simply had to): (photo)
I was just gonna suggest the same xD
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Yes vote of course. Really nice improvements.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Nahoc
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Noticed last year and noticing it again now so just wanna point it out. When you tag a DS movie with Movie # the movie length is wrong. For example Super Mario 64 DS should be 9:50.64 and not 9:48.93. Is there any way to fix that?
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Definitely Nahoc.
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Thanks for the encodes Aktan!
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
Lag Counter 60 FPS: http://dl.dropbox.com/u/102137768/LagCounter60FPS.lua
Nahoc wrote:
You need to record/playback a movie for it to work (because mupen).
Experienced Forum User, Published Author, Experienced player (648)
Joined: 5/16/2009
Posts: 235
This is a M64 Editor I made some months ago to simplify my life and avoid doing hexediting while TASing N64 games. Nahoc told me to share with you all so here it is! hehe. The tool is quite simple, after loading a .m64 file, it lets you change some of the basic data such as framecount (VI/s), description, title or rerecords. Then it lets you watch the input of each frame of a TAS file, and modify it. Finally I added a Replacement part, useful when you are working on a TAS with different files, and you want to merge them in one. I used this when building the Super Mario 64 "120 stars" TAS for example, adding single stars to the final project. x32 version: http://dl.dropbox.com/u/102137768/M64%20Editor%20x32.exe x64 version: http://dl.dropbox.com/u/102137768/M64%20Editor%20x64.exe (Your antivirus may detect this file as a virus or something, since it's an .exe file created with Visual Studio, but it's not a virus.)
1 2
6 7 8 9 10