Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
xxNKxx wrote:
Edit: Well, my idea nearly correct, I was join black screen at 4068032957 but I don't have 3 boost items. But when I try change other spots and other frames but still join black screen at 4068032957 I have same items. Now we know RNG numbers will not same on per stage and if we know how to calculator RNG numbers, we can get 3 boost items not too hard
What were the specifics of your test? Don’t forget that the items you get are different depending on your Rank, and when you unlock a new Grand Master item you only get 1 of it no matter what RNG.
Did you finish stage 2 with Hathor’s RNG number and Rank, and still got different items? That would be weird…
Edit: by the way in your Lua code, my friend said:
[19:41:46] Julian:
<<< local find_RNG = 0
[19:41:52] Julian: should be "" not 0 but lua will probably accept it anyway
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
I was rewatching this and wondered what 100% meant. Saving the game shows your %, right? Is this a 100%? Do they collect those boss medals that require you to beat them in a particular way (IIRC)? If not, then those don’t count towards 100%?
This is the YT vid description: “In this run, hellagels and red-crowned-crane collect all secret disks, life ups, BM upgrades and sub tanks, achieving a 100% completion rate.”. Is that what a 100% run is? I thought life ups didn’t matter…
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Ah, that’s what you meant. I thought you meant we didn’t have to wait 1 frame inbetween X presses.
Luck manipulation isn’t that important right now, we still need to finish stage 1 and TAS stage 2 properly.
Also I plan on disassembling the RNG to try and figure it out eventually.
Edit: omg we were actually losing 1 frame from pressing Start too late in the menu before stage 1… but the movie desyncs after that. I think lag changes depending on Start timing…
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Huh, nice, you waited at the second to last hit to get a faster last hit. Did you hex-edit the input after Katanal or did you re-TAS it?
I was thinking that maybe we should test Ayame (the girl) to see if she’s faster… I know she has a longer boss fight on stage 5 or something, but I think she has faster movement and that may help in the long run.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
feos wrote:
ALAKTORN wrote:
feos wrote:
/me fails to see the connection between loading savestates and how they should not be in a video capture.
Have you not read the topic?
Have you never encoded your TAS?
I just tried it out and got a great video capture out of an input file I recorded with a bunch of savesloads (yes, in mame-rr).
Yeah, you haven’t read the topic. synnchan isn’t talking about normal TASing (consider adelikat’s above post).
Edit: lol adelikat edited his post for some reason.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
If you’re playing back a movie file and everything isn’t exactly the same every time, it’s not a problem of RNG or whatever, it’s a problem of sync that could depend on the emulator, ROM, settings or whatever else on the hardware level rather than the software. Edit: nevermind the fact I’m misusing the hardware/software words… hope you can understand what I mean anyway.
I don’t do Dolphin TASing so I can’t be much help, but don’t bother trying to “manipulate” the RNG, if it’s not even synching in the first place.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Thanks to the helpful programming of my friend STL I found the RNG. :)
0x000C55E8, 4 bytes
xxNKxx, you can play around with it and see what you can find out if you want, I just wanted to try finding the address, I won’t be working on the run for another while I think.
Edit: something very interesting, I think the RNG is reset to 0 after each level. This means that hex editing would be super easy if the only difference is luck.
Edit2: I think it’s at least 31 bits big. Maybe 32, maybe the 32th bit is used to determine sign? I don’t know. /// Nvm calculated wrongly, 1073123057 is still only 30 bits. 1073741824 would be 31.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
dunnius wrote:
MUGG wrote:
I don't like this kind of "input end" where the character stands around for another minute before the game ends. I prefer fastest real time
I agree. It was this reason that I voted no on E.T. I didn't like the standing around at the end. It detracts from the entertainment.
What entertainment? This is a fighting game speedrun, it’s not supposed to be entertaining. Check out Cooljay’s run for entertainment.
I found it entertaining exactly for the clever trick, but I’m just trying to say that it’s weird to hear about entertainment in a fighting game speedrun aimed for the vault.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
lapogne36 wrote:
The only way I know is to use a lua script
while true do
memory.writeword(0x000CC270, 118);
memory.writeword(0x000CC274, 75);
--And so on
emu.frameadvance();
end;
However I am not sure it's what you are looking for
I still haven’t tested this (not sure if I can dump RAM Search/Watch from PSXjin in a copy-pastable manner), but can’t Lua freeze a value in place instead of needing a specific one written? I risk some terrible behaviours if I’m to set whole bunches of memory to 100 or something instead of simply freezing them to their current value.
Also an easier solution might be to edit the emulator itself to allow for mass-cheating in RAM Search? feos? :P
Edit: STL wrote this for me:
addrs = { 0x4555, 0x8346738 }
values = {}
function readFreeze()
values = {}
for addr in addrs do
value = memory.readdword(addr)
values[addr] = value
end
end
function writeFreeze()
for addr, value in pairs(values) do
memory.writedword(addr, value)
end
end
readFreeze()
while true do
writeFreeze()
emu.frameadvance()
end
Edit2: gives “:6:attempt to call table value” error, though.
Edit3: this works, but the value isn’t really frozen, the emu can still update it or something:
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Oh my god, that’s genius. :D Loved the ending.
As far as changing default options goes, I think if the game doesn’t penalize you for doing it, then sure why not. In Street Fighter IV, if you do that, your final score doesn’t get sent to the online leaderboards, so that could be an argument against it for that one game…
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
feos wrote:
Disassembling requires looking up opcode mnemonics. And ASM has little to do with what most people understand by the term "programming languages". Also, you can't really learn ASM (only basics of how it works), since it varies so much, especially on consoles. Disassembling requires knowing what to do with debugger.
Which is all stuff that becomes much easier to learn if you’re already a programmer. I guess I shouldn’t have said “requires”, but you were also wrong in saying that “it doesn’t help”.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Hurray for me, I’ve killed another TAS with my strategies.
We aren’t using an input file btw. The whole run’s botted. Also we’re stuck on M3 because we can’t figure out a pattern that’s come up in the RNG with the new strat. We need the AI to suicide a unit which is a 1 in 5829.02 chance, and that’s kind of hard to get if you can’t manipulate the RNG.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
TASeditor wrote:
ALAKTORN wrote:
If you want to be a good TASer, most definitely programming is required. Just look at the likes of Mr.Wint… that’s a good TASer.
bot coder
What makes you think the theoretical best TASer wouldn’t be anything other than a bot coder? You wouldn’t call Mr.Wint a TASer? He creates bots that make TASes, that’s TASing at the highest level.
Knowing programming languages doesn't help you in figuring out RNG
This is just wrong, though. Disassembling requires programming knowledge.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
If you want to be a good TASer, most definitely programming is required. Just look at the likes of Mr.Wint… that’s a good TASer. Still, you can be very decent without much programming knowledge, RAM Search can be learnt by someone who knows nothing about programming I guess, I’ve never known a lot about programming but I can still understand a lot of what people talk about when discussing memory and the likes. But programming and maths knowledge definitely make you a better TASer.
I think this site has guides to learn how to use RAM Search, so you should start there if you don’t know how to use it yet. For RNG like TASeditor says, its RAM address won’t help you much most of the time; to make use of it you need the aforementioned programming and/or maths knowledge. Luck manipulation is probably the hardest part of TASing.
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
I get the feeling every single player that submits a terrible TAS as his first attempt and then gets taught by posters never goes on to learn anything. Is there a counter-example of that?
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
* time attack in japanese,
Usually shortened to TA, you could try タイムアタック which is the katakana.
* "play-through" or "long-play" in japanese
実況プレイ is how I’ve seen it used on YT, that’s for playthrough. Don’t know about long play.
Nicovideo usually has everything you need to know about Japanese games…
Experienced Forum User, Published Author, Player
(99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
Dooty wrote:
Even though the speed is the same, Dixie moves faster in the air. (???)
Wut. Are you sure you’re looking at speed?
MKDasher found a way to calculate speed based on XYZ position using lua, and it seems much more precise in most games than the game’s RAM speed addresses seem to be. Maybe this is another case where that would help…?
