Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
diggidoyo wrote:
Please elaborate.
From my experience I know that sometimes even with RAM Watch and Lua scripts you may not be able to fully optimize games. Without them you never will.
RAM Search doesn’t really require programming knowledge. You just need to know a few things which you can easily learn while following a tutorial. You don’t even need to find the pointer if that’s not working for you, but then you’ll have to re-find the addresses every time you re-open Dolphin. The pointer just serves as a never-changing address that gives you the possibility for a quick math calculation to identify the addresses you really need, after you’ve found them for the first time.
If you’re ok with having to redo it every time, you could try to find Y position and RNG without bothering with pointers. There are a bunch of tutorials on the site on how to use RAM Search. I’ll just say a few quick things: you have to set a data size, 1/2/4 bytes, 1 byte is 2^8 = 256 possible values. 2 bytes is 2^16, 4 is 2^32. Say you want to find lives, and you know they go from 0 up to 99 (let’s say that’s max). That would be 1 byte, as it’s contained within 256 values. Y position and RNG are usually things that get really big so you should try 4 bytes for them. Next thing is whether the value is signed or not (you can also pick hexadecimal but that rarely helps). Signed means that it can become negative, unsigned it can’t. 1 byte signed becomes −127…0…128, not 0…255. Same logic for the other sizes. Something like Y position could be signed, going negative low on the screen and going positive higher on the screen. The RNG is probably not signed. Knowing that you can start doing searches with the given parameters. Say you want to find Y, you can do data size = 4 bytes, signed, do “equal to previous value” and search it a few times while not moving. Then you can jump, pause emulation, and do a “higher than previous value” (assuming jumping up raises your Y position). Etc., you just have to try a few things and use your brain to figure out how to identify what you’re looking for. Not sure what else to say, you should be able to use RAM Search after that.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
diggidoyo wrote:
As I said, trial and error works just the same.
No it doesn’t, and as a fellow TASer my heart breaks when I read these posts. The image you posted doesn’t show the settings so I’m not sure what you were trying to say with that. Not being able to find the 0000 address is a problem with those settings from what I’ve read, but if you’ve configured them correctly then I don’t know how to help. You could try asking in that topic.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
diggidoyo wrote:
I tried to follow the instructions here but there were no addresses that ended in 0000's and that a pointer scan revealed a value with only one offset.
RachelB wrote:
Did you do this?
In Cheat Engine, click Edit, then Settings. Go to Scan Settings and check MEM_PRIVATE, MEM_IMAGE, and MEM_MAPPED.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
I really think you guys should try to find the RNG in RAM. For such a boss fight I wouldn’t expect changing events before entering the fight will give any better RNG during the final part of the boss.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Don’t even know what I’m translating. Edit: OK I get it now, hope it’s understandable.
“Try no use savestate
すいません 英語で説明しずらいので日本語で書きます
—It’s hard to explain in English so I’ll write in Japanese.
ディスクチェンジ前にメモリカードセーブ
—Before changing the disc, I save in the memory card.
そのデータから開始してディスクチェンジ画面まで進めた
—Loading that data I was able to progress until the “disc change” screen.
フレームアドバンスを使わず エミュレータポーズもせず psx>disc controlsから
open(5分待つ) none(5分待つ) close(5分待つ) open(5分待つ) disc2クリック(5分待つ) clonseを試しましたがやはりだめでした
—Without using frame advance, nor pausing the emulator: PSX -> disc controls -> open (waited 5 minutes) -> none (waited 5 minutes) -> close (waited 5 minutes) -> open (waited 5 minutes) -> click on disc2 (waited 5 minutes) -> clonse; I tried doing this but as I thought it didn’t work.
また別のゲームでも試しました
—I’ve also tried a different game.
Gundam 0079 THE WAR FOR EARTH
—^that game
こちらはpassword再開でdisc2からできるので
—In this game it’s possible to start on disc2 by inputting a password, so I did:
disc1スタート＞パスワード入力＞discchange画面となり
—start on disc1 -> enter the password -> it becomes the “disc change” screen
同様に試しましたがダメでした
—I tried in the same way (to change the disc) but it didn’t work.”
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Timing issues should be your only problem. Seeing as you want to do a speedrun an emulator isn’t that great of an idea as it probably won’t run at the exact same speed as the console… but maybe it could. Anything else not acting as a console would be a bug, not expected behaviour. So you shouldn’t expect it to be inaccurate.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
FractalFusion wrote:
ALAKTORN wrote:
“ Uses speed/entertainment tradeoffs ”
Does that still apply? Where?
The blackout glitch is not used. This glitch is where, after beating a boss, if you press start on the frame where X starts walking toward center, the screen blacks out. It saves about 20 frames (by not having fadeout) but adversely affects the music by having it play at soft volume (like when in the submenu) until you open the submenu again.
For this reason, Hetfield90 does not use the glitch in the TAS.
…That sounds like a really cool glitch though. Sure it might get boring if it’s always used, but at least using it once could make it interesting and faster at the same time.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Submitted already? Was expecting a few more days of trying out things.
Edit: if Fractal helped so much with the run shouldn’t he be part of the authors?
Posting down my ratings so I remember when it gets published: 8/9.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
If you’re saving frames then the RNG would be shifted compared to before. Not much point going by old results, just try it all anew with the improved file?
Does Dolphin have RAM Watch? It sounds like you aren’t watching the RNG. You could maybe use Cheat Engine or something else too for it. Finding the RNG shouldn’t be too hard, I’d look for 4 bytes changing every frame.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
I’ve enconded our WiP for Time Survival. We actually have “TAS’d” up to M8, but since I found the better M3 strategy we went back, and couldn’t manipulate the RNG for it and got stuck. So the WiP is only up to M3.
https://www.youtube.com/watch?v=IM1of7jTHIE
From the description: “We don’t know how to manipulate the RNG on M3. The AI is supposed to suicide the Mech, Infantry, and Tank, and move the AA where our Tank moves on the last day for the kill.”
If you’re wondering why we enter Turn Survival at the start, that’s RNG manipulation. It’s obviously a lot slower in terms of real time, but like I’ve said before this TAS aims for the best possible in-game record, so our manipulation is for in-game time. The Neo suicide on M1 is probably a 1 in a couple thousands chance (going by memory), so there’s no way of getting it unless we manipulate the RNG before entering Time Survival (we calculated this), and it’s required for beating the map in 2 in-game seconds (otherwise you get 3 seconds).
Edit: oh and yes, it’s kind of a NG+. There’d be little point in running Time Survival from clean SRAM.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Anty-Lemon wrote:
lxx4xNx6xxl wrote:
Hey Omni? I know this might be a stupid question but, what is the P in PRNG?
Psuedo
*Pseudo.
:P
I guess that makes sense Omni. I’ve never really had to look at an RNG that deeply except for in Advance Wars: Dual Strike, and I think that one uses the full 65536 values.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
If you want to find RAM addresses you need to use RAM Search not RAM Watch. That’s where you put the addresses you want to look at after you’ve found them.
People have already told you all that’s possible to teach about it in the other topic, I’m also not sure why you made a new topic.
You’ve said there was a lot left to be filtered, well then just keep filtering. Use your brain and be smart about how you filter things and you’ll be able to find something as simple as enemy HP in no time.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
“There are about 1000 values out of the 43534”
I’m actually confused by this: why are there 43534 values total? If the RNG is 2 bytes, shouldn’t there be 65536?
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Hetfield90 wrote:
Considering how I had already killed every possible enemy after Vile and created the max number of dust sprites during down time, it would've been impossible to get it without waiting around for an amount of time larger than what the pattern would have saved.
I’m not sure why you’re equating “maximum advancements” with “the correct amount to manipulate a pattern”? Did you like test every iteration of killing and not killing every enemy etc. and still didn’t get the pattern?
Are you watching the RNG address in RAM and have you reverse-engineered its basic 1-step formula?
