Most Generation PRNG Help / Information

[mingot]

8C90885C -- Remember though that when you put this in it will skip the first three frames. Click the coin toss a few times and toss another and lets see if we can't find ourselves :)

was a dumbass and had it wrong.

And if you want to see yours here is the whole thing backed up a few calls:

156BE40B

 

[LightningFusion]

I don't have coin flip in platinum :(. I'm going to catch a pokemon, check its IV, find it, and just go between apps and see how its affected I guess.

I caught 2 pokemon 5 secs apart, but the frames and time were only 62 and 1.03 different, respectively.

 

[mingot]

Thats way way to little for the RNG to be going at 60 calls per second. I'd imagine that's battle rng calls and grass rng calls to be honest.

If you happen to have a few fossils that might be an even better test. Just dont reset :)

EDIT: Just tried myself catching two bidoof back to back and had a 32 frame spread difference. The RNG is definately sitting still for the most part. I am going to go have a smoke and capture #3 and see if there is a similar difference. Catch #3 at frame 88.

So:

Catch 1: Frame 3
Catch 2: Frame 37 (immediate catch)
Catch 3: Frame 88 (5 min smoke break)

 

[LightningFusion]

I got the coin and other app in video and trying to duplicate a similar thing.

I did the tap twice thing on the one app 100 times, switched to coin flip and flipped 5 times, and change one more. This would be frame 1206 but since it's different starting one, I am going to find the difference.

 

[mingot]

What's the "other" app in this case?

 

[LightningFusion]

It's a friendship app. Tap to show it's friendship or double tap empty space to make the pokemon jump.

I have to get some sleep and will most likely be on later tomorrow night.

You should, or I could later tonight, try doing fossil's and see how the RNG is affected, and keep track of the steps when you go outside and back in to get it. See how the RNG is affected by 2 extra steps and try a minute later on one.

 

[=Jonny=]

how can you tell the difference in frames between a catch and another ?

 

[LightningFusion]

He has the reverse seed calculator so you can see all the up coming seeds. And he catched some of them and found out where they were.

Edit: I hope mingot sees this.

I copied the the video thing where it tapped the one 85 times and went forward and flipped the coin 5 times, then went forward one.
I did the tap thing 100 times instead since 85 went to 0:00:58 from the beginning seed so I was hoping it would go to a seed I didn't already passed but the results from the egg was not in my current list. I think it went to an earlier seed in the game.

 

[Pink]

Alright, someone tell me what to do on Azelf to test this stuff. I dunno what you need me to test, so just let me know what the best course of action would be.

 

[LightningFusion]

I would say wait for mignot's third version of it to see the backward thing.

Maybe try this: Get some fossils. Save in front of fossil person who you take your fossil too. SR, get fossil, find IVs and PID, then get another one, and keep doing this, keep track of how many steps you take to see how the RNG is affected from walking, and see if it's consistant, not based on time, etc.

 

[Pink]

Well, I got a couple repeat Timid spreads on Azelf.

Timid: 140/126/81/134/76/136

I wasn't using the AR to check IVs because it screwed with the timing on the SRing. But, I was able to repeat this a couple of times, so yeah.

Timid #482 Azelf: 10 / 23 / 12 / 9 / 2 - 3 / 8 - 9

EDIT: Just got it again. Going to try getting it with the AR on.

 

[LightningFusion]

I think the most important thing we could do is find out how the RNG varies between catching pokemon.

 

[=Jonny=]

I don't know if this has been said yet but I just translated a part of the article.
It says there are two different Random Numbers
A normal random number (wild pokemon)
and an "incubation" random number (eggs)

they both start with the same initial seed, the only difference is the way the seeds are consumed. for the NRN consuption depends on the numbers of step taken etc...while for the IRN it depends ONLY on the usage of poketech apps...

each time you get on the friendship checker / make your pokemon jump by double clicking on the touch pad the seed consumption is 2*(number of pokemn in your party)
each you use the coin flip or the roulette apps the seed consumption is one

steps and elapsed time DO NOT AFFECT this number wich basically determines the pid of the egg at the day care mansion...so it's a piece of cake to adjust once one knows the initial seed.

 

[LightningFusion]

Ok that's great news. I don't know how mingot's new program works for backtracking, but we need it to go to the starting seed, so we can find out how many times to mess with the app to forward it to the breeding one we want.

Jonny - do you know what the 1026 came from in the video. I know it's his shiny pokemon but if you put the 5b15025b number in for the starting value and the ID's in, the spread doesn't come up as shiny.

 

[Pink]

Johny, can you expand on the IRN? Give me some steps and I'll go test it now, to try and get a spread to happen more than once.

Could it be that the shiny spreads and all that changed for Platinum?

EDIT: What would his shiny spreads be with his IDs Fusion?

 

[LightningFusion]

Try this, save the game in front of the egg. Turn off the screen. Change the time to whatever you can get the same result. Start the game and double tap the one screen or flip the coin, take the egg, and find out the result. Save, get a new egg, turn it off, change the time and back on and do the same thing you did with the app and take the egg.

According to mingot's program, frame 4444, 1:14:06 was the first one.

 

[=Jonny=]

first you should be able to back-determine your starting seed

 

[LightningFusion]

I know how get got the 1026 number from the one japanese app. I can confirm this later tonight or tomorrow if I can trade someone a pokemon and find out my secret ID. Also, mingot's program would have to get the inital seed, not just the one right before it.

Someone needs to find out how the RNG fluctuates after his new version of the program comes up based on steps.

I'll be on later tonight most likely.

 

[Pink]

Ok, I'm new to all this, and the IV breeding just confuses me. Especially with the seed manipulation. Can you tell me if this right?

1) Set time, get the egg, check IVs.
2) Try to recreat IVs by resetting time, and checking IVs of a new egg.
3) Once I get it, reset time, then check IVs after flipping a coin.
4) Recreate those?

The only thing is, does the amount of time before I take the egg affect the SEed and the IVs? Cause then it might be difficult to get a constant IV. Unless I'm stupid.

 

[=Jonny=]

It's pointless to do this test without knowing the pid. The IRN only affects the PID not the IVs and can be maniplated to make a shiny PID.

 

[Pink]

Ohhhh. OK. I'm sorry, I misunderstood. One of links from the youtube video had away to get the PID right?

 

[mingot]

Ok, looks like I have a lot to update the OP with, so good work guys so far.

I'll update on where I am :/

Last night I got my IVs->SEED application working and from there started working on taking that seed and some information about when you reset and allowing a hunt for the initial seed.

It's not going to hot. At this point I've either messed something up or misunderstood the forumla (which seems strange as I can use the formula and get the same sample results as the initial japanese page) OR the seeding formula has changed for the english version (unlikely, but I will test it).

Here is how I am going about trying to find the initial seed right now:

Given Nature/IV's find a list of valid pids for a monster (usually one. sometimes more) and create a list of these "monster seeds". The user then enters the following information about when they reset: Year, Month, Day, Hours, Minutes. (missing are: seconds and delay on continue screen).

Then I do this for each one of the "monster seeds":

I use this seed to go backwards through the previous seeds (a maximum of 10000 times, which is way too much, but I wanted a high number for testing) and then for each one of those seeds I do the following:

I use the formula on page one and calculate every combination with the values the user entered for Year, Month, Date, Hours, Minutes and cycle through seconds 0-59 and delays of 0-36000 looking for a match.

I keep a list of all matches along with how far they were from the "monster seed" and what the seconds and delay were that needed to be substituted. I generally get a few possible hits, but the problem is they'll have HIGH delay numbers or be offset -1000 back from the initial seed when they really should be withing a couple. I'm very frustrated at this point, but will keep messing with it.

Jonny -- If you see ANYthing about the initial seeding in the pages you are translating please let me know. As much different information to verify my method would be good.

 

[Pink]

Ouch O_o. Have fun with that Mingot. Anything we can do to help at this point?

 

[=Jonny=]

the part I'm translating right now is quite hard to understand for me. But I discovered something very interesting.
I have rs save of me in front of a zapdos in a emulator. I saved a SGM of the game and caught a zapdos found its ivs and nature and put them in the app the japanese players use for platinum, then pressed a button...
I obtained results for pid, gender, ability and another number wich I thougth it was the seed...
So put this number in RNG reporter and generated the spreads. I reloaded the SGM and moved one frame on with the special feature present on the emulator, caught another zapdos and its ivs and nature matched with the first spread generated by RNG reporter...
I know this is not platinum related but since I'm using an english version of RS this could be a confirmation that the RNG is the same for the english versions and that there actually is a method to compute the seed

 

[Wichu]

All 3rd gen+ games use the same RNG, so that would happen. However, R/S seed differently, so they're not exploitable yet. If you did it through save states, it would also save the RNG state.

And Zapdos in R/S? :S