Past Gen RNG Research

[Kaphotics]

5th Generation & Pickup

img of researcher

Pickup only works if you don't run.

Carrying 6 Pickup Mons and defeating a wild poke:
It cycles 7 times after battle if you don't get a pickup, 8 times if one gets a pickup, hence a calc proccing another calc immediately after.

Exiting Seed from the battle is considered the Initial Seed for this post:

Initial

Advanced Once, 
  if    SEED*100>>32    is < 10
     then [SEED+1]*100>>32 for Pickup Item Value (0-99)
    else, advance once and repeat for next slot.
  end
6 iterations, there is one calc afterwards for something else.

Pickup Table located here, although the placement for the 1%'s is off. Prism Scale is 98 for level 100 (leftovers is 99).

Double Pickup, Prism Scale and Max Revive

0-29 30-39 40-49 50-59 60-69 70-79 80-89 90-94 95-97 98 99
 30%   10%   10%   10%   10%   10%   10%    5%    3% 1% 1%

 

[Bond697]

you can simplify that, actually. the game does it like so:

-rng advance
-save u32 to r0
-umull u32(r0) by 0xA to r4/r5 then r0/r1(the result we want in r4 is moved to r1)
-move r1 -> r0
-cmp r0 to 0, then bcs
-go ahead with the pickup calc if the result of r0 * 0xA is 0, move on if == 1-9
-rng advance
-this calc is r0 * 0x64

to make it easier, just make it ((seed) >> 32 * 0xA) >> 32; if result == 0, do pickup calc

 

[chiizu]

On request, here's the translation of the double battle encounters found in the same pokebw readme file as the compound eyes encounter (which were implemented in PPRNG probably a month ago when this readme was first found...).

Researcher screenshots with actual RNG frames are left as an exercise for the researcher.

0:  Start
1:  Sync check (applies to both Pokemon, skip this if lead Pokemon
                has compound eyes just like in single encounter)
2:  Double battle check (as posted previously in this thread)
3:  ESV for right Pokemon
4:  Unknown (same unknown as in normal wild encounters,
             my guess is Pokerus infection check)
5:  ESV for left Pokemon
6:  Unknown (same as 4 above, I'd imagine)
7:  Right Pokemon PID
8:  Right Pokemon Nature
9:  If right Pokemon can hold an item, roll for item as previously described,
      otherwise this is skipped (i.e. no RNG value consumed)
10: Left Pokemon PID
11: Left Pokemon Nature
12: If left Pokemon can hold an item, role for item as previously described,
      otherwise this is skipped (i.e. no RNG value consumed)

If the double battle check fails, the single battle frame continues on as normal (i.e. steps 3, 4, 7, 8, and 9).

The fun take away here is that the PID and nature of the left Pokemon is affected by whether the right Pokemon can hold an item.

Also, regarding compound eyes removing the sync check, this means that some PIDs for swirling dust / bridge shadow encounters will be hittable with compound eyes when they would not be with a non-compound eyes lead.

 

[Bond697]

wordswordswordswords

Thank you.

Here's the shaking patch info, btw:

Hide

shaking patches
---------------------------

First, there's a secondary step counter at 2257034(White U) that counts 0-19 like so:

021AABE8 2000     mov     r0,#0x0   // r0 from 2257028 to 0
021AABEA 43C0     mvn     r0,r0 // set r0 to NOT 0, so r0 = !0 = FFFFFFFF
021AABEC 330C     add     r3,#0xC // 2257028 + 0xC - 2257034 - shakling patch step counter address
021AABEE 4282     cmp     r2,r0 // r2 is shaking patch step counter value, compare counter value w/ 0
021AABF0 D201     bcs     #0x21AABF6 // to 21AABF6 if r2 is 0, shaking patch calc stuff
021AABF2 1C50     add     r0,r2,1 // tack 1 onto the value of the step counter, throw sum in r0
021AABF4 6018     str     r0,[r3] // store this new value (step counter +1) to 2257034
021AABF6 4804     ldr     r0,=#0x5B8
021AABF8 681A     ldr     r2,[r3] // load the new step counter to r2
021AABFA 5A08     ldrh    r0,[r1,r0] // halfword at 232e500 to r0
021AABFC 4282     cmp     r2,r0
021AABFE D301     bcc     #0x21AAC04 // true, branch

021AAC04 2000     mov     r0,#0x0
021AAC06 4770     bx      r14


When it does hit 0, it does this to see if it should create a shaking patch:

02005728 B538     push    {r3-r5,r14} // this all just sets up the rng for multiplication
0200572A 1C05     mov     r5,r0
0200572C 480A     ldr     r0,=#0x20AA1B4
0200572E 6904     ldr     r4,[r0,#0x10]
02005730 68A0     ldr     r0,[r4,#0x8]
02005732 68E1     ldr     r1,[r4,#0xC]
02005734 6822     ldr     r2,[r4]
02005736 6863     ldr     r3,[r4,#0x4]
02005738 F096EC98 blx     #0x209C06C

(rng.multiply())(via 64-bit multiply function)

0200573C 6922     ldr     r2,[r4,#0x10] // put the rng back, get ready to do the standard u32 * XXX(or skip it for pid calcs)
0200573E 6963     ldr     r3,[r4,#0x14]
02005740 1810     add     r0,r2,r0
02005742 414B     adc     r3,r1
02005744 6020     str     r0,[r4]
02005746 6063     str     r3,[r4,#0x4]
02005748 1C18     mov     r0,r3
0200574A 2100     mov     r1,#0x0
0200574C 2300     mov     r3,#0x0
0200574E 1C2A     mov     r2,r5
02005750 F096EC8C blx     #0x209C06C

u32 * XXX multiplier- 3E8 in this case(1000dec)
rnd.calc();

02005754 1C08         mov  r0, r1   // final result
02005756 BD38         pop  (r3-r5,r15)  // restore stack, jump back to restored r15

021AA8D4 490C     ldr     r1,=#0x5BA
021AA8D6 5A62     ldrh    r2,[r4,r1] // r2 == 0xA
021AA8D8 210A     mov     r1,#0xA // r1 == 0xA
021AA8DA 4351     mul     r1,r2 // r1 == 100dec
021AA8DC 4288     cmp     r0,r1 // compare result of rng*1000dec mult w/ 1000(r0-r1)
021AA8DE D304     bcc     #0x21AA8EA // carry isn't clear unless it's less than 100, so branch if < 100

if >= 100
-------------
021AA8E0 1C30     mov     r0,r6 // false? set r0 to 2257028- the rng advancement routine
021AA8E2 F000F993 bl      #0x21AAC0C // done, off to other stuff

I'll put the rest up later with some notes.

e: not that this is super-important, but npc movement is (u32 * 4) >> 32. npcs move based on a timed coundown and once it hits 0 the rng is called and that calculation is done. i never bothered to figure out what direction 0-3 correspond to, though, exactly.

[21:50] <Kaphotics> 0 - Up, 1 - Down, 2 - Left, 3 - Right

there, directions

 

[Kaphotics]

Wondercards with a locked-in gender but variable nature have their own method of PID generation separate from the normal generation.

Test Data

0	E9F11E5F06414BB4	Initial Pre-Load
Everything is loaded from the card from 1 to 23
24	C459BD1164D3FAEC	Initial: "OT" received a "PKM" (Pre-IV)
25	DD993995A584E9DF	27	HP
26	0244C0E5E7D83ABE	0	Atk
27	0BF80A6EF31779B9	1	Def
28	41FA83D8D89CA5C0	8	SpA
29	6C81B76ED2E7C383	13	SpD
30	D7F97C17D9CCDC72	26	Spe
31	8814F98BAF1099BD	
32	417D45ACE4246B54	
33	EA3491B7DF40EAE7	
34	157F7C07B2C9EAE6	
35	D193AF140CCF6181	
36	000427771FAD1FA8	0 (hardy) We look Forward to your next visit	
37	F99EEE51E2FD040B	End Talk-Double Advance
38	EB0A5CBEB5701A1A

Up/T Low/S
1892 0089
181B 0000
XOR = 0

Another Trial

0	9E2FE436D6F3BA08	Initial Pre-Load
24	41FA83D8D89CA5C0	Initial Pre-IV
25	6C81B76ED2E7C383	13	HP
26	D7F97C17D9CCDC72	26	Atk
27	8814F98BAF1099BD	17	Def
28	417D45ACE4246B54	6	SpA
29	EA3491B7DF40EAE7	29	SpD
30	157F7C07B2C9EAE6	2	Spe
31	D193AF140CCF6181	
32	000427771FAD1FA8	
33	F99EEE51E2FD040B	
34	EB0A5CBEB5701A1A	
35	22D151AC420ED505	
36	D11A3A8EEF2356BC	20 (calm)
37	15245718F1A672EF	End Talk-Double Advance
38	BDCEA77E62DCDE0E	

Up/T Low/S
18E8 00F3
181B 0000
XOR = 0

"GLAN" Wondercard Generation Structure

0 - Initial
(card loaded/pre-data, dependent on card)
25 - HP IV (>>27)   [offset varies]
26 - Atk IV (>>27)
27 - Def IV (>>27)
28 - SpA IV (>>27)
29 - SpD IV (>>27)
30 - Spe IV (>>27)
31-35 PID Gen
36 - Nature (*25,>>32)

This was the Hydreigon present. Didn't test the Golurk.

Essentially, there's 2 more dummy frames before the IVs and 1 more in the PID routine. Should be easy to adapt this method in!

 

[Bond697]

white forest pids match gift pokemon as far as i can see. there might be a different formula for it, but i've not done the disassembly yet.

 

[chiizu]

This appears to be the main method used to detect whether the game is running on a DS or DSi. I base that statement on the fact that it's reading a value from a register that doesn't exist on the DS and checking for a specific result. On a DS, this should return a 0 value if no such register exists.

This function is called in lots of places and the result is used to decide which code or values to use in numerous functions.

Code

------------------- DSi check  (White JP)
-- determines whether running on DSi, and stores result to 02151008 (1 = DSi, 0 = DS)
--- check is done by looking at a register 0x04004000, which doesn't exist on DS
--- if lowest two bits == 0x1, *02151008 = 1
--- subsequent calls just read from 02151008 rather than checking the register
:02085C60 E59F003C ldr r0,[r15, #+0x3c]      ; r0 = 0x02151004 (baes address of a global variable structure)
:02085C64 E590001C ldr r0,[r0, #+0x1c]       ; r0 = r0[1c](*02151020) - static bool for onetime check
:02085C68 E3500000 cmp r0,#0x0               ; if (r0 == 0) - if not called before
:02085C6C 1A000009 bne 02085C98              ; {
:02085C70 E59F0030 ldr r0,[r15, #+0x30]      ;   r0 = 0x04004000
:02085C74 E3A02001 mov  r2,#0x1              ;   r2 = 1
:02085C78 E5D00000 ldrb r0,[r0, #+0x0]       ;   r0 = *04004000 (DSi register?)
:02085C7C E3A01001 mov  r1,#0x1              ;   r1 = 1
:02085C80 E2000003 and  r0,r0,#0x3           ;   r0 = r0 & 0x3
:02085C84 E3500001 cmp r0,#0x1               ;   if (r0 != 0x1) - check that low 2 bits of register are 01
:02085C88 E59F0014 ldr r0,[r15, #+0x14]      ;   r0 = 0x02151004
:02085C8C 13A02000 movne  r2,#0x0            ;     r2 = 0 (not 0x1, so not DSi)
:02085C90 E5802004 str r2,[r0, #+0x4]        ;   *02151008 = r2 (store in global)
:02085C94 E580101C str r1,[r0, #+0x1c]       ;   *02151020 = r1 (set flag to indicate result is cached
:02085C98 E59F0004 ldr r0,[r15, #+0x4]       ; r0 = 0x02151004
:02085C9C E5900004 ldr r0,[r0, #+0x4]        ; r0 = *02151008  (read cached result)
:02085CA0 E12FFF1E bx r14

 

[Kaphotics]

Somewhat unrelated to Pokemon Generation, but can influence it via frame advancement:

NPC actions...

The RAM area associated to NPC movement is in the 0x02250000 region, and is different for each location. Tethered NPCs (don't step) have...

A timer set by the RNG to determine when to call for the next move
A Movement Limiter (Limits Movements determined by RNG to a specified range)
A NPC-# to instruct which NPC to move
A NPC directional to instruct it how to move.

As we've seen already, the NPC directional is determined by *4. However, if the NPC cannot move in that direction a call is still consumed to reseed the timer.

Nimbasa Pokemon Center Moving NPC RAM view

For moving NPCs, -0x10 from the face directional to have the move directional.
Below these are the movement limits and so on. Meh.

There's a small cooldown timer for moving and facing so that they won't be allowed to move even if the timer goes down (moving too much).


tl;dr -- even if NPCs don't move they still can consume frames

 

[Kaphotics]

RSFRLG Egg Inheritance
nailed it

Sapphire Memory Locations

03004818 	RNG Seed

03004360	Party 1 PID
030043C4	Party 2 PID
03004428	Party 3 PID
0300448C	Party 4 PID
030044F0	Party 5 PID
03004554	Party 6 PID
020286d0	Daycare A PID
02028720	Daycare B PID
020287E8	Egg LowPID
020287EA	Egg Stepcounter

Sapphire Egg Inheritance Test

=====================================
38310586	Seed
28575 frames back gives D661

Ditto 		1/1/1/1/1/1
Sceptile	0/0/0/0/0/0
Ditto deposited first 		= A
Sceptile deposited second 	= B
=====================================
IV's, Method 1 Frame, PIDs, and Inheritance
-----------------------------------
Pokemon's IV's -- 0/2/1/0/30/11	
appears as Method 1 Frame 5
PID: 918869601 (dec)	36C4[I]D661[/I] hex

B/2/A/B/30/11 is what I got.

B/x/A/B/x/x on Frame 9 for HGSS
-----------------------------------
Pokemon's IVs -- 31/1/29/0/13/0	
appears as Method 1 Frame 7
PID: 2024527457 (dec)	78AB[I]D661[/I] (hex)

31/A/29/B/13/B is what I got.

x/A/x/B/x/B on Frame 11 for HGSS
-----------------------------------
Pokemon's IVs -- 10/0/0/1/1/3	
appears as Method 1 Frame 9
PID: 3055081057	(dec)	B618[I]D661[/I] (hex)

10/B/B/A/1/3 is what I got.

x/B/B/A/x/x on Frame 13 for HGSS
-----------------------------------
14/0/12/28/1/0	Frame 13
PID: 2689717857 (dec)	A051[I]D661[/I] (hex)

14/B/12/28/A/B is what I got.

x/B/x/x/A/B on Frame 17 for HGSS

An example...

[I]Lower Half[/I] of the PID comes from [I]Method 1[/I] when the egg becomes available.
--
[I]Upper Half[/I] of the PID comes from [I]Method 1[/I] Frame 1. 
IV's come from [I]Method 1[/I] Frame 2.
Inheritance comes from [I]HGSS's Inheritance[/I] Frame 6.

==

Contrary to what was thought, eggs are not generated on the stepcounter's reset. Freezing the "resetted" step counter to an arbitrary value still allows for eggs to be generated.

Hide

Freezing the total step counters will prevent eggs from being generated, located at -0x8 and -0x10 of the 0-255 stepcounter location.

 

[GoldenBanana]

Thank you so much, Kaphotics, for your RSFRLG findings! I was just beginning to research this, only to find that you've almost uncovered all that there is to know.

I am a little confused about the information you've presented though. Firstly, I don't understand where you get D661 from (that's supposed to be the Sapphire version seed, right?) and its corresponding frame. When I use NeoSyrex's FR/LG seed finder application, I get 5a0, which is the seed you're stuck with when the game tells you that the battery has run dry. I also get a different frame, and what's more is that I can only find the spreads you listed when I enter 5a0 as the seed in RNG Reporter.

Secondly, what is the long number you've listed underneath the IVs, and what is the number that is followed by _D661? What is their significance?

Lastly, what is a uPID, and what is it used for?

Let me see if I've got the gist of your findings:
Say I'm on some seed, frame 1287. The uPID (whatever that is) will come from M1 frame 1287, the IVs from M1 frame 1288, and the inheritance from Breeding (HGSS) frame 1292?

 

[Kaphotics]

Oops, I changed my shorthand to something more coherent (re-read the post :X).

As OmegaDonut noted earlier about the egg PID generation method in RSFRLG, the egg PID is generated in two steps. All I did during my period of research was replicate what he did (independently, mapping memory locations as I went), and figured out the last part that we didn't know: inheritance.

I wasn't around / researching Gen 3/4 things when they were popular, so I have no clue how HGSS/Emerald inheritance works, I just know that's what's happening for inheritance. We know Nintendo likes to re-use stuff, but whodathunk HGSS's inheritance was reused from RSFRLG?

===

For my case, the D661 was the lower part of the resulting egg PIDs (I rerolled save states at different frames). It probably was from [00005A0's frame 85456 in Researcher]. I saved my game and restarted just to make sure that the value stayed there.
If you were to use the initial seed 000005A0 in RNG Reporter, the frames I hit for inheritance were starting at 16835. For my tests, I just used the "current seed", which is just a further along frame.

The RNG can be thought of as a number line, you can start from many points and still end up at the same place. :)

===

You got the gist of it :)

 

[Kaphotics]

B/W Wild Pokemon Levels

Encounter Levels for everything but aquatic encounters are set between the same two values. For fishing/surfing(+spots) they are not the same, thus resulting in the level range obtainable.

The formula is this for the "Level Calc" Frame:

Load the  LowLimit on the Pokemon's Level (BaseLVL)
Load the HighLimit on the Pokemon's Level (HighLVL)

{{{ (u32*100>>32);  
                   % [[HighLVL - BaseLVL + 1]] }}} 
                                                   + BaseLVL

In simple terms:

Take the *100 of the seed
         You have a number (0 thru 99)
Mod it by the (level difference +1)
         You have a number (0 thru gap+1),
         Modding results in a # (0 thru gap)
Add the resulting # to the lowest possible level. 
         You now have the wild Pokemon's Level

For non-aquatic encounters this still occurs, the mod(levelgap+1=1) always yields 0, and thus you get the static level.
Chiizu's post on double battle structure has an "Unknown" frame consumed, this is what is happening there. For every wild encounter, the Level is calculated right after ESV.


See the Encounter Slots: Water post for examples and frame occurrence.

 

[GoldenBanana]

More on RSFRLG egg generation

I don't know if this has been figured out already, but I'll post it anyway. Also, I apologize in advance for any illiteracy I may show on this subject.

It seems that when the number at the location in memory that records the Egg Step Counter ends in FF (like 1FF or 2FF), there is a chance that the old man will have an egg. Since you can't check the game's memory on a cartridge, the only way to properly do this would be to place the parents in the daycare (rather than collecting an egg with the parents already in the daycare), walk 254 steps 1 square away from the old man, and save. Walking one step towards the man will trigger a chance for the egg to be generated. Every 256 steps afterward will trigger this same chance.

It's also important to note that the appearance of the egg doesn't appear to be seed dependent. Resetting on the same seed (5a0) and walking one step to advance the egg step counter to FF did not always result in an egg being generated.

Also, for people who are interested in doing this on a cartridge, I found that the time it takes to generate the egg from the moment you press a direction (basically the walk animation) is 16 frames.
----------------------
I'm looking at the RNG Seed in Ruby's memory, and as I'm advancing the frame one-by-one manually, some method 1 frames seem to be skipped. Is anyone else experiencing this problem?

 

[chiizu]

Starting PID Frames
or why this guy's table and code works as a substitute for the real (giant) table and strangely complicated code.

The simplified version of the PID start frame calc follows. Take note of the array named kakuritu ('percentage' in Japanese).

Taken from the same guy's program

Note: I don't see any difference between the behavior of simulateOffset from this guy's program and what RNG Reporter is doing, despite the comment in the RNG Reporter source that this code is broken. I see the difference now.

void simulateOffset()
{
    int kaisu = 5;  //多くの場合は5回だと思いますが絶対に5回とは言い切れません
    for(int i = 0; i < kaisu;i++)
    {
        selectMissionAll();
    }
}

int selectMission(int pattern) {
    static const int kakuritu[6][5] ={  //謎の確率テーブル
      {  50,100,100,100,100 },
      {  50, 50,100,100,100 },
      {  30, 50,100,100,100 },
      {  25, 30, 50,100,100 },
      {  20, 25, 33, 50,100 },
      { 100,100,100,100,100 }
    } ;
    
    if(pattern > 6)return 0;
    if(pattern < 0)return 0;
    int rn = 0;
    for(int i = 0;i < 5;i++)
    {
        if(kakuritu[pattern][i] == 100)return i;    //確率が100だったときは乱数を発生させずに処理を抜けます
        rn = ransuu.genRand64(101);                    //0～100の乱数を発生させます。
        if(rn <= kakuritu[pattern][i])return i;        //乱数が確率の値以下だったときは、その場で処理から抜けます。
    }
    return 0;
}

int selectMissionAll()
{
    for(int i = 0;i < 6;i++)
        selectMission(i);  //上のselectMissionの全パターン分を順番に呼び出します。
    return 0;
}

The table and code he gives is a simplified version of the 224 row in-game table and the functions that use it as part of a process that ends up consuming the initial RNG frames in the game. The purpose of that process is currently unclear, but the Japanese developer may be thinking that it has something to do with Entralink missions and power levels (judging by the function names and some comments on his twitter).

The actual table the game uses (note that names are just my choices based on how a variable is used in this context)

struct TableRow
{
  byte  index;
  byte  sumCheck;
  byte  percent;
  byte  evenOrOdd;
};

struct Table
{
    TableRow  row[224];
};

The raw bytes of the rows (all numbers in hexidecimal), located at 02176ACC in White JP
01 30 19 04
01 30 19 03
01 26 19 04
01 26 19 03
01 20 19 04
01 20 19 03
01 18 19 04
01 18 19 03
01 11 21 04
01 11 21 03
01 0e 21 04
01 0e 21 03
01 0b 21 04
01 0b 21 03
01 08 32 04
01 08 32 03
01 05 32 04
01 05 32 03
01 02 32 04
01 02 32 03
01 00 32 04
01 00 32 03
01 00 32 04
01 00 32 03
01 00 64 04
01 00 64 03
02 32 19 04
02 32 19 03
02 28 19 04
02 28 19 03
02 22 19 04
02 22 19 03
02 19 1e 04
02 19 1e 03
02 13 1e 04
02 13 1e 03
02 10 1e 04
02 10 1e 03
02 0d 1e 04
02 0d 1e 03
02 0a 32 04
02 0a 32 03
02 07 32 04
02 07 32 03
02 04 32 04
02 04 32 03
02 00 1e 04
02 00 1e 03
02 00 32 04
02 00 32 03
02 00 64 04
02 00 64 03
00 2a 14 04
00 2a 14 03
00 22 14 04
00 22 14 03
00 1a 19 04
00 1a 19 03
00 12 19 04
00 12 19 03
00 0a 21 04
00 0a 21 03
00 05 21 04
00 05 21 03
00 00 32 04
00 00 32 03
00 00 64 04
00 00 64 03
05 19 21 04
05 19 21 03
05 0f 32 04
05 0f 32 03
05 05 32 04
05 05 32 03
05 00 64 04
05 00 64 03
04 30 0f 04
04 30 0f 03
04 30 14 04
04 30 14 03
04 30 19 04
04 30 19 03
04 2b 0f 04
04 2b 0f 03
04 2b 14 04
04 2b 14 03
04 2b 19 04
04 2b 19 03
04 26 0f 04
04 26 0f 03
04 26 14 04
04 26 14 03
04 26 19 04
04 26 19 03
04 21 0f 04
04 21 0f 03
04 21 14 04
04 21 14 03
04 21 19 04
04 21 19 03
04 1c 0f 04
04 1c 0f 03
04 1c 14 04
04 1c 14 03
04 1c 19 04
04 1c 19 03
04 17 14 04
04 17 14 03
04 17 19 04
04 17 19 03
04 17 1e 04
04 17 1e 03
04 12 14 04
04 12 14 03
04 12 19 04
04 12 19 03
04 12 1e 04
04 12 1e 03
04 0d 14 04
04 0d 14 03
04 0d 19 04
04 0d 19 03
04 0d 1e 04
04 0d 1e 03
04 08 14 04
04 08 14 03
04 08 19 04
04 08 19 03
04 08 1e 04
04 08 1e 03
04 04 14 04
04 04 14 03
04 04 19 04
04 04 19 03
04 04 1e 04
04 04 1e 03
04 04 32 04
04 04 32 03
04 00 14 04
04 00 14 03
04 00 19 04
04 00 19 03
04 00 21 04
04 00 21 03
04 00 32 04
04 00 32 03
04 00 64 04
04 00 64 03
03 36 0f 04
03 36 0f 03
03 36 14 04
03 36 14 03
03 36 19 04
03 36 19 03
03 36 1e 04
03 36 1e 03
03 31 0f 04
03 31 0f 03
03 31 14 04
03 31 14 03
03 31 21 04
03 31 21 03
03 2c 0f 04
03 2c 0f 03
03 2c 14 04
03 2c 14 03
03 2c 21 04
03 2c 21 03
03 27 0f 04
03 27 0f 03
03 27 14 04
03 27 14 03
03 27 21 04
03 27 21 03
03 22 0f 04
03 22 0f 03
03 22 14 04
03 22 14 03
03 22 21 04
03 22 21 03
03 1d 14 04
03 1d 14 03
03 1d 1e 04
03 1d 1e 03
03 1d 32 04
03 1d 32 03
03 18 14 04
03 18 14 03
03 18 1e 04
03 18 1e 03
03 18 32 04
03 18 32 03
03 13 14 04
03 13 14 03
03 13 1e 04
03 13 1e 03
03 13 32 04
03 13 32 03
03 0e 14 04
03 0e 14 03
03 0e 1e 04
03 0e 1e 03
03 0e 32 04
03 0e 32 03
03 08 14 04
03 08 14 03
03 08 1e 04
03 08 1e 03
03 08 32 04
03 08 32 03
03 03 14 04
03 03 14 03
03 03 1e 04
03 03 1e 03
03 03 32 04
03 03 32 03
03 00 19 04
03 00 19 03
03 00 1e 04
03 00 1e 03
03 00 32 04
03 00 32 03
03 00 64 04
03 00 64 03

The inner functions (decompilation of actual function used in game, outermost function missing)

r6 = ?
Address = memory address of some data structure

This small loop is around 02176240 in White JP
for (OuterCounter = 0; OuterCounter < 6; ++ OuterCounter)
{
  *(Address + 6 + OuterCounter) = r6
  f(Address, OuterCounter)
}

This function is located at 2176254 in White JP
f(uint8_t *Address, OuterCounter)
{
  Sum = ((uint16_t*)Address)[6] + ((uint16_t*)Address)[7];
  i = 0;
  Table  *table = &BigTable;  // address of above 224 row table
  
  do
  {
    if ((table->row[i].index == OuterCounter) &&
        (table->row[i].sumCheck <= Sum) &&
        IsOdd(table->row[i].evenOrOdd))
    {
      percent = table->row[i].percent;
      if ((percent == 100) || (percent >= NextRNG(101))
      {
        Address[OuterCounter] = i
        return
      }
    }
  }
  while (++i != 224)
  .
  .
  .
}

The important thing that makes the smaller table work is that at start up, everything located at Address in the above code is 0, and r6 seems to always be 0 as well. This means that the sum of Address[6] and Address[7] is always 0, and therefore only the rows of the giant table that have a sumCheck value of 0 along with an evenOrOdd value that is odd (3 for this table) will be actually considered.

If you filter out all rows with a sumCheck value greater than 0 and a evenOrOdd value of 4, and then reorder the rows by the index byte you get the following:

Hide

Index 0 Rows
00 00 32 03    // % = 50
00 00 64 03    // % = 100

Index 1 Rows
01 00 32 03    // % = 50
01 00 32 03    // % = 50
01 00 64 03    // % = 100

Index 2 Rows
02 00 1e 03    // % = 30
02 00 32 03    // % = 50
02 00 64 03    // % = 100

Index 3 Rows
03 00 19 03    // % = 25
03 00 1e 03    // % = 30
03 00 32 03    // % = 50
03 00 64 03    // % = 100

Index 4 Rows
04 00 14 03    // % = 20
04 00 19 03    // % = 25
04 00 21 03    // % = 33
04 00 32 03    // % = 50
04 00 64 03    // % = 100

Index 5 Rows
05 00 64 03    // % = 100

Add additional 100% rows to each index so that all have 5 rows (to match index 4) and what you will see is the content of the simplified table in code given at the beginning.

Given that the real table contains lots of rows that would be considered when the sum value is larger (if r6 is non-zero, for example), as well as having duplicate rows with alternating values of 3 and 4 in the final byte, it's fairly easy to conclude that this table serves some larger purpose in the game beyond a seemingly random initial RNG frame advancement.

I don't expect this post will lead to new discoveries (though who knows what may come up down the line - this table definitely has other uses), but I thought it was nice to have actual evidence of where the simplified table being used in RNG Reporter (derived originally by the guy linked to at the beginning) ultimately comes from.

 

[Kaphotics]

Black and White Swarms
Basically the same as the Wild structure, but with a calculation in between Sync and ESV

[I]if[/I]         {{  u32*100>>32  }} < 40
   [I]then[/I]    return active swarm, and still consume ESV
   [I]else[/I]    calculate ESV, standard Pokemon (not swarm)

Swarm Level Range: (15 thru 55)

0	Initial
1	Sync
2	Swarm
3	ESV
4	Level
5	PID
6	Nature

Most notably, Swarm Pokemon don't have any held items thus they don't have a calculation for them at the end of the call cycle.
Essentially the Swarm Calculation is the same thing as the doubles grass calculation.

Example at Route 10 - Tyrogue

0	C06688275F850487	Initial
1	9F089A75F2BEA706	Sync
2	36A8487BE5EABA21	Swarm
3	009490E25644B6C8	ESV
4	03D6575C2E8BC3AB	Level
5	21A3141CA6AE543A	PID
6	9EA717AFD396E3A5	Nature

 

[Kaphotics]

scribbly post, gen 5

frozen seed C9DAE91DEB4AAAED generates an egg in less than 20 steps for a "They Get Along" breed.

Trying to find out why...
0 C9DAE91DEB4AAAED
1 02F00960728FE344 02F00960... (top4bits= 0)

It's definitely not the main step counter. Probably that 17steps/rng one.

scribblesection gen 3

Outputting SIDs for Emerald ID abuse

0-59 60-79 80-99
60% 20% 20%


Old Rod
Slot 0: 0-59 60%
Slot 1: 60-89 30%
Slot 2: 90-94 5%
Slot 3: 95-98 4%
Slot 4: 99 1%

Slots 0234 and 1 make 70 30
Need to determine if it's not 0-69/70-99.

Good Rod
60/20/20



Super Rod
Slot 0: 0-39 40%
Slot 1: 40-79 40%
Slot 2: 80-94 15%
Slot 3: 95-98 4%
Slot 4: 99 1%

034 1 2 make 45/40/15
023 1 4 make 59/40/1

Feebas and the Level Calc in Emerald hardly WIP
how does this work I dont even

What I think it will be: 
Before ESV, Low bit determines if Feebas will appear. Works like sync check.


Fishing Levels are determined by u16%(max-min+1)+min
Can't get it to work for Method 2 and Feebas. wat???????????

examples

Can't even figure this out. sjghbdriekgjndreghbeargszgrzredgdehbdsxrenh

with help from zari to get test results

 

[Kaphotics]

RSEFRLG Encounter Slots

RNG Seed is 32 bits, only uses u16 as we know already.

Backwards Method:

1. Find the frame the IVs occur on for the initial seed (000005A0).
2. Take the PID of that frame and %25 (0x19) to get Nature.
3. Subtract One from the frame, determine if it is odd or even.
4. Set up researcher, as described in the example (until a method is available).
   • I call it Method H, it's VERY similar to Method J / K in gen 4 due to how it works. It exists before gen 4, thus H is before J.
5. Work backwards from the PID frame, only looking for your Nature in custom 2 in the even/odd frames (You determined even or odd in step 3).
6. Once you have found it, two frames back on Custom 3 is the ESV.

Instead if you predict it forwards, you can possibly have a few different encounter slots for the PIDIV, just like Gen 4.

Slot      0     1     2     3     4     5     6     7     8     9  10  11
Rate    20%   20%   10%   10%   10%   10%    5%    5%    4%    4%  1%  1%
Value  0-19 20-39 40-49 50-59 60-69 70-79 80-84 85-89 90-93 94-97  98  99

Encounter Slots for Gen 3 games can be found at poccil's web site.
[URL="http://upokecenter.com/games/rs/guides/"]RSE[/URL] and [URL="http://www.upokecenter.com/games/fireleaf/firelocs.php"]FR[/URL]/[URL="http://www.upokecenter.com/games/fireleaf/leaflocs.php"]LG[/URL]

RS Example

LOCATION: [URL="http://bulbapedia.bulbagarden.net/wiki/New_Mauville#Pok.C3.A9mon"]New Mauville Basement[/URL]
Initial Seed: 00005A0 (Dead Battery)
Sweet Scenting Gloom

Emerald Example

LOCATION: [URL="http://bulbapedia.bulbagarden.net/wiki/New_Mauville#Pok.C3.A9mon"]New Mauville Basement[/URL]
Initial Seed: 00000000 (Emerald)
Sweet Scenting Gloom

FRLG example

LOCATION: [URL="http://bulbapedia.bulbagarden.net/wiki/Cerulean_Cave#B1F_2"]Cerulean Cave B1F[/URL]
Initial Seed: 00005542
Version: Fire Red
Sweet Scenting Sceptile (lulz)

Do note that this does make some spreads unobtainable for this encounter type just like in J/K.
Method 1 based generation will have the same excluded spreads as Method K, but for Method 2 and 4 based generation, nobody knows yet!

Emerald Sync
Essentially the same as in method K, in which the Nature call's u16 low bit determines yes-0 or no-1. If no, consume another frame for nature.

Emerald Sync Example

Much thanks to FractalFusion, poccil and mingot :)

 

[Kaphotics]

Daily Item RNG

Hide

Twist Mountain Fossil Man in B/W

(u32*7) >> 32

yields 0-6, follows this #-Item distribution:
0    Root  Fossil (Lileep)
1    Claw  Fossil (Anorith)
2    Helix Fossil (Omanyte)
3    Dome  Fossil (Kabuto)
4    Old   Amber  (Aerodactyl)
5    Armor Fossil (Shieldon)
6    Skull Fossil (Cranidos)

Fish of the Day

(u32*21) >> 32

yields 0-20, follows this #-Fish distribution:
0	Carvanha
1	Basculin
2	Sharpedo
3	Feebas
4	Milotic
5	Goldeen
6	Seaking
7	Poliwag
8	Poliwhirl
9	Poliwrath
10	Krabby
11	Shellder
12	Luvdisc
13	Kingler
14	Cloyster
15	Finneon
16	Horsea
17	Seadra
18	Qwilfish
19	Lumineon
20	Kingdra

Treasure Hunter's Daily Item

(u32*27) >> 32

yields 0-26, follow this #-Item Distribution
0	Blue Flute
1	Yellow Flute
2	Red Flute
3	Black Flute
4	White Flute
5	Shoal Salt
6	Shoal Shell
7	Big Pearl
8	Heart Scale
9	DeepSeaTooth
10	DeepSeaScale
11	Lucky Egg
12	Metal Coat
13	Dragon Scale
14	Up-Grade
15	Lucky Punch
16	Metal Powder
17	Thick Club
18	Stick
19	Protector
20	Electirizer
21	Reaper Cloth
22	Magmarizer
23	Razor Fang
24	Razor Claw
25	King's Rock
26	Dubious Disc

Lost Item at Anville Town

Hide

(u32*10) >> 32

Pay attention to: 022598D0 english white
Value is loaded when it displays what item you receive.
When you dismiss his final dialogue (happy item), 
           next week's item value will be there.
Depending on Block Shuffling, it is stored at 2032A, quantity at 2032C.

You have to win 7*x battles in the battle subway for him to give you the next item.
You will get 1*x of that item when you talk to him. Max of 10.
So, you can get up to 10 Rare Candies each day.

0	Rare Candy
1	Full Restore
2	Max Revive
3	Elixir
4	Protein
5	Iron
6	Calcium
7	Zinc
8	Carbos
9	HP Up

Item Codes: Reference
0017 	Full Restore	1
001D	Max Revive	2
0028	Elixir		3
002D	HP Up		9
002E	Protein		4
002F	Iron		5
0030	Carbos		8
0031	Calcium		6
0032	Rare Candy	0
0034	Zinc		7

 

[Kaphotics]

Sports Dome Trainers

The game generates 6 trainers (order of position 654321 in image) out of 52 in a Pool to choose from.

Research Scribbly important mechanics

Mid - Research, probably needs breaking.

Seed: B760BD8432B5DA3E
Frame	Trainer
1	49
2	42
3	4E/4D
4	3A
5	3C/3D
6	5C/5D
7	5A/59
8	47/46
9	41/40
10	65/64
11	56/57
12	4F
13	68
14	4D

u32*100	7	10	18	22	30	36	43	
trainer	3A	3C/3D	40/41	42	46/47	49	4D/4E	

u32*100	45	47	62	67	71	87	95
trainer	4D	4F	56/57	59/5A	5C/5D	64/65	68

As you can see, it likes to follow the linear scale set from the Upper32 seed.

Trainer order is located at 0x2033E, depending on shuffling. Occupies a 12 digit span (00TR * 6)

Non Random Trainer Hexes

00 = Ungenerated
01 = Smasher Elena (30s)
02 = Smasher Elena (40s)
03 = Smasher Elena (65s)
04 = Smasher Aspen (40s)
05 = Smasher Aspen (65s)
06 = Smasher Mari (65s)
07 = Infielder Alex (20s)
08 = Infielder Alex (30s)
09 = Infielder Alex (40s)
0A = Infielder Alex (65s)
0B = Infielder Connor (40s)
0C = Infielder Connor (65s)
0D = Infielder Todd (65s)
0E = Striker Tony (20s)
0F = Striker Tony (30s)
10 = Striker Tony (40s)
11 = Striker Tony (65s)
12 = Striker Roberto (40s)
13 = Striker Roberto (65s)
14 = Striker Marco (65s)
15 = Linebacker Dan (20s)
16 = Linebacker Dan (30s)
17 = Linebacker Dan (40s)
18 = Linebacker Dan (65s)
19 = Linebacker Bob (40s)
1A = Linebacker Bob (65s)
1B = Linebacker Jonah (65s)
1C = Hoopster Bobby (20s)
1D = Hoopster Bobby (30s)
1E = Hoopster Bobby (40s)
1F = Hoopster Bobby (65s)
20 = Hoopster John (40s)
21 = Hoopster John (65s)
22 = Hoopster Lamarcus (65s)

23 = Backers Hawk & Dar (30s)
24 = Backers Hawk & Dar (40s)
25 = Backers Hawk & Dar (65s)
26 = Backers Cam & Abby (65s)
27 = Backers Ami & Eira (30s)
28 = Backers Ami & Eira (40s)
29 = Backers Ami & Eira (65s)
2A = Backers Alf & Fred (65s)
2B = Backers Kay & Ali (30s) 
2C = Backers Kay & Ali (40s) 
2D = Backers Kay & Ali (65s) 
2E = Backers Les & Web (65s) 
2F = Backers Joe & Ross (30s)
30 = Backers Joe & Ross (40s)
31 = Backers Joe & Ross (65s)
32 = Backers Ai & Ciel (65s)
33 = Backers Masa & Yas (30s)
34 = Backers Masa & Yas (40s) 
35 = Backers Masa & Yas (65s)
36 = Backers Kat & Phae (65s)

37 = Clerk F Wren
38 = Waiter Clint
39 = Waitress Bonita
3A = Ace Trainer Charlie
3B = Ace Trainer Lucille
3C = Lass Sophie
3D = Rich Boy Anthony
3E = Policeman Jeff
3F = Black Belt Lao
40 = Harlequin Charley
41 = Artist Pierre
42 = Scientist Simon
43 = Scientist Blythe
44 = Psychic Gerard
45 = Psychic Madhu
46 = Worker Matthew
47 = Worker Tyler
48 = Gentleman Renaud
49 = School Kid Alan 
4A = School Kid Sally
4B = Roughneck Fletcher
4C = Janitor Caleb
4D = Pokefan Colin
4E = Pokefan Darcy
4F = Dancer Davey
50 = Youngster Kevin
51 = Fisherman Devon
52 = Depot Agent Josh
53 = Doctor Jules
54 = Nurse Kirsten
55 = Pilot Leonard
56 = Backpacker Alexander
57 = Backpacker Patty
58 = Battle Girl Janie
59 = Parasol Lady Gwyneth
5A = Clerk M Nelson
5B = Clerk M Fredric
5C = Baker Lilly
5D = Veteran Arlen
5E = Veteran Sayuri
5F = Nursery Aide Leah
60 = PKMN Breeder Owen
61 = PKMN Breeder Brooke
62 = PKMN Ranger Alain
63 = PKMN Ranger Heidi
64 = Socialite Emilia
65 = Lass Dana
66 = Musician Boris
67 = Maid Alicia
68 = Hiker Russel
69 = Preeschooler Evan
6A = Preeschooler Mia

6B = YOURSELF (unused)

More exist beyond, but they are not part of the random trainers.

6C = PKMN Ranger Mylene, who is in Desert Resort

.

sup arlen
223BF0A houses all 6 trainer #'s, is saved to save file.

see next page for how to calc.

Process 6 times total to get 6 random Trainers.

0 = Clerk F Wren
[U]1 = Waiter Clint[/U]  -- Fresh Water
[U]2 = Waitress Bonita[/U]  -- Fresh Water
[B][U]3 = Ace Trainer Charlie[/U][/B] - PP Up
[U][B]4 = Ace Trainer Lucille[/B][/U] - PP Up
[U]5 = Lass Sophie[/U]  -- Ultra Ball
6 = Rich Boy Anthony
7 = Policeman Jeff
8 = Black Belt Lao
9 = Harlequin Charley
10 = Artist Pierre
[U]11 = Scientist Simon[/U]  -- Ether
[U]12 = Scientist Blythe[/U]  -- Ether
13 = Psychic Gerard
14 = Psychic Madhu
15 = Worker Matthew
16 = Worker Tyler
[u]17 = Gentleman Renaud[/U]  -- Nugget
18 = School Kid Alan 
19 = School Kid Sally
20 = Roughneck Fletcher
21 = Janitor Caleb
22 = Pokefan Colin
23 = Pokefan Darcy
24 = Dancer Davey
[U]25 = Youngster Kevin[/U]  -- Ultra Ball
26 = Fisherman Devon
27 = Depot Agent Josh
[U]28 = Doctor Jules[/U]  -- Max Revive
[U]29 = Nurse Kirsten[/U]  -- Full Restore
30 = Pilot Leonard
[U]31 = Backpacker Alexander[/U]  -- Repel
[U]32 = Backpacker Patty[/U]  -- Repel
33 = Battle Girl Janie
34 = Parasol Lady Gwyneth
35 = Clerk M Nelson
36 = Clerk M Fredric
[U]37 = Baker Lilly[/U]  -- Moomoo Milk
[B][U]38 = Veteran Arlen[/U][/B] -- PP Max
[B][U]39 = Veteran Sayuri[/U][/B] -- PP Max
[U]40 = Nursery Aide Leah[/U]  -- Rare Candy
41 = PKMN Breeder Owen
42 = PKMN Breeder Brooke
43 = PKMN Ranger Alain
44 = PKMN Ranger Heidi
[u]45 = Socialite Emilia[/U]  -- Big Pearl
46 = Lass Dana
47 = Musician Boris
48 = Maid Alicia
49 = Hiker Russel
50 = Preeschooler Evan
51 = Preeschooler Mia

 

[Bond697]

u32*51 >> 32

Process 6 times total to get 6 random Trainers.
There's still a little iffyness (nonrepeating and items?)

considering that u32 * XX >> 32 is less than half the formula(and 51 is wrong), i'm not surprised there's "a little iffyness".

 

[Kaphotics]

RSFRLG Encounters (Part 2)
Part crossposted from FRLG thread.

If you freeze a value to the main seed location, it will act as the initial seed for PID generation, thus bypassing any advancement from sweet scent (even though it doesn't "consume"). A little hackish, but gets research done.

Now that I can successfully test individual seeds (knowing what Nature I'll hit) via seed freezing in the emulator, I can verify those ESV->Slots.

Fishing Encounter Slots

Here's what I believe them to be:

Old Rod
Slot 0: 0-69 
Slot 1: 70-99 

Good Rod
Slot 0: 0-59
Slot 1: 60-79
Slot 2: 80-99

Super Rod 
Slot 0: 0-39
Slot 1: 40-79
Slot 2: 80-94
Slot 3: 95-98
Slot 4: 99

Battle Decision Trigger Calcs

R/S: Upper16 %0xB40 < 16*Density
	Restated: 	if (u16%2880)/16 < D	, pass encounter

Modifications (Poccil's Site):    
   1. If the Mach Bike or Acro Bike is used, multiply by 4/5.
   2. If the Black Flute had been used, multiply by 1/2. If the White Flute had been used, multiply by 3/2.
   3. If the first Pokémon in party has a Cleanse Tag, multiply by 2/3.
   4. If the first Pokémon in party has Stench, multiply by 1/2. If the first Pokémon in party has Illuminate, multiply by 2. (In either case, ignore if the Pokémon holds a Cleanse Tag.)

Maximize: (*6) White Flute, Illuminate
Minimize: (*1/20) Stench, Black Flute, Bike
Apply to the resulting value to density.


FRLG: See [URL="http://upokecenter.com/games/rs/guides/notes.php?section=Encounters"]Poccil's site[/URL] for psuedocode.
Emerald: See [URL="http://upokecenter.com/games/rs/guides/notes.php?section=Encounters"]Poccil's site[/URL] for psuedocode.

PKM Generation Routine (Method H), dumbed down

Setup of everything before PID:

0 - Initial
1 - Battle Decision,		 If Sweet Scent, pass		 = Battle
2 - Encounter Slot Value (0-99), via Upper16 %100		 = Slot
3 - Encounter Level (from Slot), via Upper16 %(High-Low+1) + Low = Level
4 - Encounter Nature, 		 via Upper16 %25		 = Nature

Now to generate a PID.

set n = 0
loop until break to get a PID%25 = Nature
5+n - Temp  Low16PID
6+n - Temp High16PID

	if High16PID_Low16PID % 25 = Nature
		set as the PID, break
	    else n=n+2
		repeat loop
	end

end

Then comes the IVs, which depend on the Method of Generation.
Not sure if this is that calc at frame 1, the rejection of IV bits, or something else.

	if Method 1
		a = 0 
		b = 0
	elseif Method 2
		a = 1
		b = 0
	elseif Method 4
		a = 0
		b = 1
	end

7+n + a   - IVSet1
8+n + a+b - IVSet2

Now this assumes method 3 is never used. I don't think I've seen it, ever. Jussayin'

=============================

I looked back at some of the spreads that Zari hit for the Feebas. On a different route (Outside New Mauville) I got a Method 1 instead of Method 4 spread.

Gen Seed: DCD4AA7E (frame 39035 hardy).
I'm gonna try the same route. If someone can give me a .sav of Sapphire on a Feebas tile, I can try to figure out how those work.

 

[Kaphotics]

R/S (FRLG?) Egg Availability [and Generation]

Egg Decision:
if 2nd Total Stepcounter == xxFF

hex (u16*0x64) / 0xFFFF   = E
dec (u16*100)  /  65535   = E

if E < Breed#, then generate egg.

Breed#
Different OT,      Same Species -- 70
     Same OT,      Same Species -- 50
Different OT, Different Species -- 50
     Same OT, Different Species -- 20

0 - Initial 
1 - ????
2 - Egg Decision
3 - Egg PID Frame
     \--->    Make Sure 0000 nor FFFF happen:

[u]Seed 	 = u16  ==> LPID[/u]
761A62EA = 7FFF ==> 8000    [+1]
5B883D42 = FFFB ==> FFFC    [+1]
506A1C76 = FFFC ==> FFFD    [+1]
F5D0CAD9 = FFFD ==> FFFE    [+1]
476DCAD9 = FFFE ==> 0001    [+3]
409E279F = FFFF ==> 0002    [+3]
923B279F = 0000 ==> 0001    [+1]
403ED126 = 0001 ==> 0002    [+1]
91DBD126 = 0002 ==> 0003    [+1]

If u16 (pre-addition) > FFFD, 
  then +3 and %0xFFFF (or -0xFFFF)
 else +1
end

It doesn't like FFFF or 0000. FFFF is just an odd case ¯\_(ツ)_/¯

RAM Addresses (Sapphire)

There's 2 large step counters (that equal each other) 
        and one small, the second large step counter is for Egg Generation.[code][COLOR="Silver"]0x020287E0  --  Nonimportant Step counter[/COLOR]
[B]0x020287E4  --  Egg Creation step counter[/B]
[COLOR="silver"]0x020287EA  --  Egg Hatch step counter[/COLOR]

You'd deposit, walk 254 steps, then save your game. Both locations are saved when you save your game, so essentially you can take just one step and talk to the man to confirm, as we already know.

Everstone does not work in RSFRLG, as it is an Emerald Only feature. You can use nature as a frame confirmation tool; have fun hatching!
I'd assume the same calculation is used to determine egg availability in Emerald and FRLG.

===

For the breeding method (UPID + IVs + Inheritance), the first call is ?????????? (second is UPID). I think it's probably Nidoran/etc (species) decision, then it generates the rest of the PIDIV as I've already described on the last page. Most likely via %100 -> 50/50. Who cares about breeding these two species in Gen 3 anyways?

 

[Bond697]

Ruby/Sapphire Feebas Encounters

There is no frame, there is no check; Feebas tiles just follow changed encounter slots.

Hide

08084AA6 0400     lsl     r0,r0,#0x10
08084AA8 0C00     lsr     r0,r0,#0x10
08084AAA 2164     mov     r1,#0x64
08084AAC F15CF9F4 bl      #0x81E0E98


081E0908 2301     mov     r3,#0x1		// r3 = 1
081E090A 2900     cmp     r1,#0x0		// 6 v. 0
081E090C D05E     beq     #0x81E09CC	// no branch
081E090E D500     bpl     #0x81E0912	// branch
081E0910 4249     neg     r1,r1-skipped
081E0912 B410     push    {r4}			// r4 to the stack(3007E00)
081E0914 B401     push    {r0}			// r0 to the stack(8A72)
081E0916 2800     cmp     r0,#0x0		
081E0918 D500     bpl     #0x81E091C	true, branch
081E091A 4240     neg     r0,r0-skipped
081E091C 4288     cmp     r0,r1			
081E091E D34F     bcc     #0x81E09C0
081E0920 2401     mov     r4,#0x1
081E0922 0724     lsl     r4,r4,#0x1C
081E0924 42A1     cmp     r1,r4
081E0926 D204     bcs     #0x81E0932
081E0928 4281     cmp     r1,r0
081E092A D202     bcs     #0x81E0932
081E092C 0109     lsl     r1,r1,#0x4
081E092E 011B     lsl     r3,r3,#0x4
081E0930 E7F8     b       #0x81E0924
081E0932 00E4     lsl     r4,r4,#0x3
081E0934 42A1     cmp     r1,r4
081E0936 D204     bcs     #0x81E0942
081E0938 4281     cmp     r1,r0
081E093A D202     bcs     #0x81E0942
081E093C 0049     lsl     r1,r1,#0x1
081E093E 005B     lsl     r3,r3,#0x1
081E0940 E7F8     b       #0x81E0934
081E0942 2200     mov     r2,#0x0
081E0944 4288     cmp     r0,r1
081E0946 D300     bcc     #0x81E094A
081E0948 1A40     sub     r0,r0,r1
081E094A 084C     lsr     r4,r1,#0x1
081E094C 42A0     cmp     r0,r4
081E094E D305     bcc     #0x81E095C
081E0950 1B00     sub     r0,r0,r4
081E0952 469C     mov     r12,r3
081E0954 2401     mov     r4,#0x1
081E0956 41E3     ror     r3,r4
081E0958 431A     orr     r2,r3
081E095A 4663     mov     r3,r12
081E095C 088C     lsr     r4,r1,#0x2
081E095E 42A0     cmp     r0,r4
081E0960 D305     bcc     #0x81E096E
081E0962 1B00     sub     r0,r0,r4
081E0964 469C     mov     r12,r3
081E0966 2402     mov     r4,#0x2
081E0968 41E3     ror     r3,r4
081E096A 431A     orr     r2,r3
081E096C 4663     mov     r3,r12
081E096E 08CC     lsr     r4,r1,#0x3
081E0970 42A0     cmp     r0,r4
081E0972 D305     bcc     #0x81E0980
081E0974 1B00     sub     r0,r0,r4
081E0976 469C     mov     r12,r3
081E0978 2403     mov     r4,#0x3
081E097A 41E3     ror     r3,r4
081E097C 431A     orr     r2,r3
081E097E 4663     mov     r3,r12
081E0980 469C     mov     r12,r3
081E0982 2800     cmp     r0,#0x0
081E0984 D003     beq     #0x81E098E
081E0986 091B     lsr     r3,r3,#0x4
081E0988 D001     beq     #0x81E098E
081E098A 0909     lsr     r1,r1,#0x4
081E098C E7D9     b       #0x81E0942
081E098E 240E     mov     r4,#0xE
081E0990 0724     lsl     r4,r4,#0x1C
081E0992 4022     and     r2,r4
081E0994 D014     beq     #0x81E09C0
081E0996 4663     mov     r3,r12
081E0998 2403     mov     r4,#0x3
081E099A 41E3     ror     r3,r4
081E099C 421A     tst     r2,r3
081E099E D001     beq     #0x81E09A4
081E09A0 08CC     lsr     r4,r1,#0x3
081E09A2 1900     add     r0,r0,r4
081E09A4 4663     mov     r3,r12
081E09A6 2402     mov     r4,#0x2
081E09A8 41E3     ror     r3,r4
081E09AA 421A     tst     r2,r3
081E09AC D001     beq     #0x81E09B2
081E09AE 088C     lsr     r4,r1,#0x2
081E09B0 1900     add     r0,r0,r4
081E09B2 4663     mov     r3,r12
081E09B4 2401     mov     r4,#0x1
081E09B6 41E3     ror     r3,r4
081E09B8 421A     tst     r2,r3
081E09BA D001     beq     #0x81E09C0
081E09BC 084C     lsr     r4,r1,#0x1
081E09BE 1900     add     r0,r0,r4
081E09C0 BC10     pop     {r4}
081E09C2 2C00     cmp     r4,#0x0
081E09C4 D500     bpl     #0x81E09C8
081E09C6 4240     neg     r0,r0
081E09C8 BC10     pop     {r4}
081E09CA 46F7     mov     r15,r14


08084AB0 0400     lsl     r0,r0,#0x10
08084AB2 0C00     lsr     r0,r0,#0x10
[b]08084AB4 2831     cmp     r0,#0x31[/b]
08084AB6 D907     bls     #0x8084AC8
08084AB8 E03E     b       #0x8084B38

I really don't feel like commenting a bunch of assembly I understand anyway, so here's a summation of what's going on:

ushort_t slots[2] = [349, 318};
ushort_t slot = 0;

feebas = (seed >> 16) % 100;

if (feebas < 50 )
{
  slot = slots[1];
  return slot;
}
else
{
  slot = slots[2];
  return slot;
}

1. fishing junk, i'm sure
2. changed encounter slot
3. level
4. nature, et al.


and just as one last note, i saw the same routine in emerald. it's late now, so i'm not checking it out, but i will tomorrow.

 

[Kaphotics]

Rock Smash in Gen 3

Uses the battle decision:
R/S: Upper16 %0xB40 < 16*Density

0 - Initial
1 - ?????
2 - Battle Decision
3 - Encounter Slot
4 - Level (uses same calc)
5 - Nature

 

[Kaphotics]

Headbutt RNG

n - 3     initial
n - 2     ????
n - 1     Slot
n         Nature
n +2x     LPID
n +2x+1   UPID
n +2x+2   IV1
n +2x+3   IV2
etc

Nobody has mapped the slots, but I can get Shroomish@30% at ~35 (Level 3). The usual /656.
Do note Bulbapedia has the wrong level distributions (I get level 3 shroomishes, which shouldn't be possible :P)

my guess

slot 0:    0-9   (10%)
slot 1:  10-19   (10%)
slot 2:  20-29   (10%)
slot 3:  30-39   (10%)
slot 4:  40-49   (10%)
slot 5:  50-59   (10%)
slot 6:  60-69   (10%)
slot 7:  70-79   (10%)
slot 8:  80-89   (10%)
slot 9:  90-94    (5%)
slot 10: 95-99    (5%)