Tell me if you want to chip in some of the work. I've pretty much got it covered, but in case you want to help, I just want to make sure we're not both going for the same ones.Language: lua--Start to guild. local waypointsx={18,18,21,21} local waypointsy={36,26,26,25} --Guild to field. local waypointsx={21,21,18,18} local waypointsy={25,26,26,37} --Tower to Bandit Cave. local waypointsx={12,19,19,22,22,46,46,42,42,45,45,43,43,44,44,42,42,38,38} local waypointsy={11,11, 7, 7,25,25,39,39,43,43,45,45,50,50,56,56,58,58,52} --First floor of Bandit Cave. local waypointsx={16,16,18,18,26,26,30,30} local waypointsy={13,14,14, 8, 8,21,21,23} --Second floor of Bandit Cave. local waypointsx={18,18, 8, 8} local waypointsy={36,51,51,50} --Third floor of Bandit Cave (to P-FROG). local waypointsx={61,61,48,48} local waypointsy={54,53,53,43} --(Base Town to field checked manually.) --Tower to Castle Armor. local waypointsx={12,19,19,22,22,46,46,37,37,44} local waypointsy={11,11, 7, 7,22,22,17,17,13,13} --First floor of Castle Armor (no NPC manipulation). local waypointsx={15,15,17,17,16,16,20,20,23,23} local waypointsy={30,22,22,17,17,13,13,11,11, 8} --Second floor of Castle Armor. local waypointsx={53,53,50,50,49,49,46,46} local waypointsy={16,14,14,22,22,29,29,28} --Third floor of Castle Armor. local waypointsx={45,45} local waypointsy={50,43} --(Base Town to field checked manually.) --Tower to Castle Sword. local waypointsx={12,19,19,21,21, 1, 1, 2, 2, 7, 7,14,14,18,18,27,27,26} local waypointsy={11,11, 7, 7,21,21,36,36,41,41,53,53,57,57,61,61,44,44} --First floor of Castle Sword. local waypointsx={12,12,18,18,16,16,11,11,14,14,12,12} local waypointsy={32,29,29,21,21,12,12,16,16,21,21,19} --Second floor of Castle Sword (no NPC manipulation). local waypointsx={36,36,38,38,40,40,36,36} local waypointsy={12,11,11, 6, 6,23,23,21} --Third floor of Castle Sword. local waypointsx={53,53} local waypointsy={17,10} --Outside WHITKEY room to Mileille's door. local waypointsx={24,24,22,22,52,52} local waypointsy={12,13,13, 6, 6, 2} --Test pattern. Use just outside tower, zigzags right and down, then marches around perimeter of enclosure. local waypointsx={12,13,13,14,14,15,15,19,19, 7, 7,19} local waypointsy={11,11,12,12,13,13,14,14, 5, 5,14,14}
MR? ... I have a habit of renaming functions and forgetting to include the rename... MR is short for Memory Read. That was my train of thought, anyway, and I'd rather not have to retype the entire function name every time.
The fact your values are "off by one" for C353 is just fine, actually. Monsters are put in one of three slots, and even if the other slots are empty, the game still rolls for those slots. If the monster was placed in slot 1, the game's first roll is given to that monster, whereas if the monster was in slot 3, the third roll is given instead. It's not really visible which one of three slots the monster is in from the battle screen, though.
The BATTLE sword, in my plans, are never given to the GAZE mutant in the first place -- She uses the HAMMER versus SEI-RYU. The spare POWER mutant gets the BATTLE sword.
I was using the STONE book against BYAK-KO in my Agility test -- What else would I use? Have you tried putting the Arts & Crafts mutant (STONE book wielder. Just in case you haven't figured out what I mean by art by now) in slot 1 directly? There's no avoiding placing the mutant in slot 1 eventually.
I'm not sure if that extra battle in W1 is so "free". The TELEPOR route requires that you switch twice to pick up the various Mana boosts. There's also the fact that TELEPOR is given to the third member, and you most certainly must swap the spare POWER mutant at least once to get that and POWER. Three swaps are required here, with three menu transitions likely.
Not eating the extra swaps, we can spare a Mana+5 for a Mana+4, or even Mana+3 if we're giving GAZE to the starter MUTANT F. Even if we are still restricted to a Mana gain, we have more possible values to use.
In Matlab, I try to keep my variables at or very near the top of the code for easy changing if necessary, but Lua insists that functions come first, so most of the good stuff is in the middle. There are only a few lines you need to be actively concerned with. First is in the function "condition". This is the condition that is true if the luck was manipulated such that we got the result we wanted. For example, if you're looking for C30B = 23 while C33B = 161, you'd alter the line defining "reportit" accordingly. I have it set up so that we can check up to three addresses or their corresponding RNG rolls simultaneously (I figure that's the most we'll ever need to manipulate, though it can be very easily expanded). The other variables you might want to change come after the "Boilerplate" section. The variables "address1", "address2", and "address3" are the ones passed on to the condition function. Unfortunately, you have to scroll back and forth between the function and those variables, coordinating their addresses with the condition you set. I'm sorry I don't know how to keep them tidier-- I tried to at least keep them as close as I could to each other. Next is "rngofinterest". This is the address that's reported to you in the summary of the results. I may make it so that it can report multiple addresses or the fractional offset, but since the condition is already verified, what we're really interested in is the frame number at which to reset. After that come the "waypoint" variables. I encoded them as two vectors, but they should really be interpreted as a 2-by-n array. The first vector contains the x-coordinates of each waypoint while the second vector contains the corresponding y-coordinates. For the default example above, the player starts at coordinates (12,11), then moves one step right to (13,11), then one step down to (13,12), and so on. As you can see, moving more than one step at a time is supported, but you need to make sure that only the x or y coordinate (not both!) changes from element to element. The program might still work when changing "rooms", which would mean the waypoints vectors would break this pattern, but I haven't yet tested it. Along with the waypoint vectors comes "wpnum", the waypoint number, which is an index to the above vectors. This number corresponds to the last waypoint visited, including where the player is currently standing (if applicable). For example, if the player is at (19,10) according to the default vectors, wpnum should be something like 8, which corresponds to the coordinates (19,14). Choose a wpnum that isn't orthogonal to the next waypoint and the program doesn't do anything. Also, you can set wpnum to 0 if you haven't reached the first waypoint. Finally, there are the "encounterrate", "framesduration", and "maxlength" variables. These are fairly self-explanatory. I've set four conditions for halting the program and returning the results: 1) The next step will trigger a battle. (Can't very well save and manipulate luck if we're in the middle of a battle, hm?) Because "encounterrate" is just one value (I wasn't able to find a RAM address for terrain), the program will only work perfectly if we don't move from one terrain to another. I figure that exceptions will be rare and can be checked easily enough individually. 2) Too many frames have passed. I usually set framesduration to 1000 for the sake of sanity, but if you want to manipulate the hell out of luck, set it to something ridiculous like 100000. 3) We've reached the end of the waypoints. No point attempting to manipulate luck if there's nowhere else to move. 4) We've obtained enough results. This is especially useful if we want just one result and we don't care when we get it. For example, I'm currently climbing the Floating Castle and if I get a +5 Mana bonus in the first slot with one expected SABERCAT anywhere along the climb, I don't need to manipulate luck any further. In that case, I would set maxlength to 1. You may be wondering about whether this program works. I'm pretty sure it does. I've checked most important aspects of it and tested it in the field. You can too. The default numbers correspond to the player standing at the base of the tower. Just step outside the Base Town and run the script. The player should walk in a zigzag pattern down and right, then walk along the perimeter of the enclosure (if there isn't a random encounter in the meantime). Go ahead and run that and tell me how it works for you. Toy around with new waypoints and even use it in the glider or on the bike! You're my beta tester. If you have any questions, comments, or suggestions to make the script better, I'd be happy oblige. Edit: I updated my code slightly. It now reports the x position, y position, and last waypoint reached. You can pass these values on to a new script I wrote:Language: lualocal function any(vector,value) local match=false for i=1, #vector do match=(match or vector[i]==value) end return match end --Apparently Lua doesn't have a built-in sign function... local function sign(x) if x>0 then return 1 elseif x<0 then return -1 elseif x==0 then return 0 else return "error" end end local function makerngtable() n={[0]=3,6,7,13,11,1,15,10,12,9,8,2,4,14,0,5} r={[0]=10,11,1,12,5,9,6,3,15,11,2,7,2,0,15,15} s={[0]=5,15,13,9,5,1,13,11,1,3,3,7,11,15,3,9} local RNGTable={} for c=0,15 do for i=0,7 do currrow=(r[c]+i*s[c])%16 currnum=n[c]+32*i RNGTable[16*currrow+c]=currnum end for j=0,7 do currrow=(r[c]+8+j*s[c])%16 currnum=255-n[c]-32*j RNGTable[16*currrow+c]=currnum end end return RNGTable end local function printcolumn(startx,starty,data) for i = 1,#data do gui.text(startx, 9*i + starty + 1, data[i]) end end local function setcourse(waypointsx,waypointsy,wpnum) local index=wpnum+1 local x=memory.readbyte(0xCCC9) local y=memory.readbyte(0xCCC8) local lr = sign(waypointsx[index]-x) local ud = sign(waypointsy[index]-y) if lr==1 and ud==0 then go = {right=1} elseif lr==-1 and ud==0 then go = {left=1} elseif lr==0 and ud==1 then go = {down=1} elseif lr==0 and ud==-1 then go = {up=1} else go = {} end return go end --This was tested to be the fastest method for saving. Hopefully, it holds everywhere. local function pauseandsave() for i = 0,1 do joypad.set(1,{start=1}) vba.frameadvance() end for i = 2,13 do vba.frameadvance() end joypad.set(1,{down=1}) vba.frameadvance() joypad.set(1,{down=1, A=1}) vba.frameadvance() joypad.set(1,{A=1}) vba.frameadvance() for i = 17,18 do vba.frameadvance() end for i = 19,20 do joypad.set(1,{A=1}) vba.frameadvance() end end --Increments wpnum only when the next waypoint is reached local function incwpnum(waypointsx,waypointsy,wpnum) local x=memory.readbyte(0xCCC9) local y=memory.readbyte(0xCCC8) if waypointsx[wpnum+1]==x and waypointsy[wpnum+1]==y then wpnum=wpnum+1 end return wpnum end local function reset() for i = 1,2 do joypad.set(1, {A = 1, B = 1, select = 1, start = 1}) vba.frameadvance() end end local function idleframes(n) for i = 1,n do vba.frameadvance() end end --Walk in a given direction until ready to save. local function walk(go) local speed=memory.readbyte(0xC5B2) local framesperstep=16/speed local startcoords=memory.readbyte(0xCCC8)+memory.readbyte(0xCCC9) while (memory.readbyte(0xCCC8)+memory.readbyte(0xCCC9))==startcoords do joypad.set(1,go) vba.frameadvance() end for i=1, framesperstep-2 do joypad.set(1,go) vba.frameadvance() end end local function condition(address1,address2,address3,RNGTable) local value1 = memory.readbyte(address1) local value2 = memory.readbyte(address2) local value3 = memory.readbyte(address3) local roll1=RNGTable[value1] local roll2=RNGTable[value2] local roll3=RNGTable[value3] --This is the super-duper important condition. Fiddle with this to test various conditions. local reportit=(any({23,137,203,161,230,243},value1) and roll3<64) return reportit end --Boilerplate... local firstrng = 0xC300 local nextbonus = 0xC30B local canrun = 0xC323 local abslot = 0xC36F local enctrgrp = 0xC343 local stepstobattle = 0xC33B local numenemies = 0xC353 local RNGTable=makerngtable() local state1=savestate.create() local state2=savestate.create() local state3=savestate.create() local rngvals={} local framevals={} local xcoords={} local ycoords={} local wpnumlist={} local enumerate={} local ind=1 movie.rerecordcounting(false) --Addresses to be luck manipulated. I figure we won't have to manipulate more than three at a time. local address1 = nextbonus local address2 = stepstobattle local address3 = numenemies --Decides what RAM address will be output in the end. local rngofinterest = nextbonus --Waypoints to be traversed. Because the player moves orthogonally, they should be staggered so only the x or y coordinate changes with each index *except* when the room changes. --(Might be possible to take into account the room-- byte CCC7-- but it's a little complicated.) local waypointsx={12,13,13,14,14,15,15,19,19, 7, 7,19} local waypointsy={11,11,12,12,13,13,14,14, 5, 5,14,14} --The waypoint number. Set to the index of the current waypoint or, if the first waypoint has yet to be traversed, just set it equal to 0. local wpnum=1 --The encounter rate and the duration in frames over which to manipulate luck. local encounterrate=4 local framesduration=1000 local maxlength=5 local startframe=vba.framecount() local nextbattlerng=memory.readbyte(0xC33B) savestate.save(state1) while (not (RNGTable[nextbattlerng]<encounterrate)) and ((vba.framecount()-startframe)<framesduration) and (wpnum<#waypointsx) and (#rngvals<maxlength) do savestate.save(state2) pauseandsave() for i=1,16 do savestate.save(state3) currframe=vba.framecount()%1000 currx=memory.readbyte(0xCCC9) curry=memory.readbyte(0xCCC8) reset() idleframes(5) if condition(address1,address2,address3,RNGTable) then rngvals[ind]=memory.readbyte(rngofinterest) framevals[ind]=currframe xcoords[ind]=currx ycoords[ind]=curry wpnumlist[ind]=wpnum enumerate[ind]=ind ind=ind+1 end savestate.load(state3) vba.frameadvance() end savestate.load(state2) local go = setcourse(waypointsx,waypointsy,wpnum) walk(go) wpnum = incwpnum(waypointsx,waypointsy,wpnum) nextbattlerng=memory.readbyte(0xC33B) end savestate.load(state1) gui.text(1,1,"#") gui.text(20,1,"RNG") gui.text(40,1,"Frame") gui.text(70,1,"x") gui.text(90,1,"y") gui.text(110,1,"Waypoint") printcolumn(1,1,enumerate) printcolumn(20,1,rngvals) printcolumn(40,1,framevals) printcolumn(70,1,xcoords) printcolumn(90,1,ycoords) printcolumn(110,1,wpnumlist) emu.pause()
I've omitted the subroutines (or whatever you want to call them) but left their names in. They should be exactly the same ones from the first Lua script in this post. This script is pretty simple-- it just executes the specified save and reset from the previous script, then checks that the result was correct. I really wanted to make it so that you could select a particular result and execute it all in one script (so we don't have to juggle values between scripts), but I apparently don't have the know-how. Perhaps you could be of assistance? I think I've got 90% of the necessary code already. Edit 2: I screwed up my example in fiddling with my program. It should be fixed now.Language: lualocal function any(vector,value) local function sign(x) local function makerngtable() local function setcourse(waypointsx,waypointsy,wpnum) local function pauseandsave() local function incwpnum(waypointsx,waypointsy,wpnum) local function reset() local function idleframes(n) local function walk(go) local function condition(address1,address2,address3,RNGTable) --Boilerplate... local firstrng = 0xC300 local nextbonus = 0xC30B local canrun = 0xC323 local abslot = 0xC36F local enctrgrp = 0xC343 local stepstobattle = 0xC33B local numenemies = 0xC353 local RNGTable=makerngtable() local ind=1 --Addresses to be luck manipulated. I figure we won't have to manipulate more than three at a time. local address1 = nextbonus local address2 = stepstobattle local address3 = numenemies --Waypoints to be traversed. Because the player moves orthogonally, they should be staggered so only the x or y coordinate changes with each index *except* when the room changes. --(Might be possible to take into account the room-- byte CCC7-- but it's a little complicated.) local waypointsx={24,24,22,22,52,52} local waypointsy={12,13,13, 6, 6, 2} --The starting waypoint number. Set to the index of the current waypoint or, if the first waypoint has yet to be traversed, just set it equal to 0. local wpnumstart=2 --The four variables that indicate when to reset. These should be obtained from the luck manipulation bot. local frame=60647 local wpnumfinish=3 local x=22 local y=10 local wpnum=wpnumstart while wpnum<wpnumfinish do go = setcourse(waypointsx,waypointsy,wpnum) walk(go) wpnum = incwpnum(waypointsx,waypointsy,wpnum) end while math.abs(memory.readbyte(0xCCC9)-x)+math.abs(memory.readbyte(0xCCC8)-y)>0 do go = setcourse(waypointsx,waypointsy,wpnum) walk(go) end pauseandsave() while vba.framecount()<frame do vba.frameadvance() end reset() idleframes(62) for i=1,2 do joypad.set(1,{A = 1}) vba.frameadvance() end idleframes(10) if condition(address1,address2,address3,RNGTable) then gui.text(1,1,"Success!") else gui.text(1,1,"Failure?") end vba.pause()
There is one chief reason for killing MUTANT M at SEI-RYU: The first party member will only go first against a higher speed enemy if the fourth member has lower Agility than the first member. If I don't give GAZE to the starter MUTANT F, I'm stuck with a slightly slower secondary MUTANT F. The only member that would be slower than this one would be the MUTANT M. When a party member dies, they go straight to the fourth slot. If someone with too high an agility is dead and thus forever stuck in slot 4 (reviving takes too long), this kills whatever chances I have of ever going first against the SABERCATs, BYAK-KO, and ATOM ANTs. The outcome is rather bleak for the survival of our heroes. The only real advantage in having two F POWER users instead giving one to the M is so I can kill the M off in this fashion.
Those two are just stand-alone functions. Stick them together under one file, call the "trigger" known as ScanReset2() every frame, and gui.text the variable str somewhere.
The plan for the Monster Trio is to hand the BATTLE sword to the spare POWER mutant. The GAZE mutant isn't using it, is she? On the other hand, I can't remember if even the BATTLE sword is enough...
The RNG is affected by what the CPU is doing internally. Generally, the music itself has a significant impact on the RNG. So does many different actions like exiting the menu. Part of the reason why I was considering using the Sound Test, of all things, to manipulate luck.
One test: Party of mutants, F F M F and see how the random battles go from there. At SEI-RYU, the M must die (two Fs will have POWER and would steal his HAMMER away from him). At monster trio, the fodder F must die. After BYAK-KO, anything goes after we get our manly man.
The third test: Party of mutants, F M F F. Swap first with fourth so the GAZE user is the one with the extra Agility. What you've been doing, with several mistakes.
--This function finds the "unknown fractional amount" that offsets the RNGs. Useful for debugging.
local function getb()
b=0
for i=0,63 do
b=b + (memory.readbyte(0xC300+i)-memory.readbyte(0xC300)-math.floor(9/64*i))%256
end
return b
endProjected| Actual |Difference
---------+---------+----------
41/64 |254 53/64|-1 52/64
16/64 |255 28/64|-52/64
9 35/64 |7 27/64 |-2 12/64
3 5/64 |2 17/64 |-52/64
3 58/64 |1 50/64 |-2 8/64Button| RNG
------+---------
None |217 10/64
A |217 10/64
B |217 10/64
Up |217 8/64
Down |217 8/64
Left |217 16/64
Right |217 15/64
Select|217 22/64
Start |217 23/64Button| RNG
------+---------
None |243 12/64
A |243 12/64
B |243 12/64
Up |243 12/64
Down |243 12/64
Left |243 18/64
Right |243 17/64
Select|243 26/64
Start |243 26/64Filling out the rest of the stuff. I hope it helps out in planning when to reset for encounters.
Step counts. Counting only the ones that increment the appropriate RNG.
Also, suppose some path takes 210 steps to finally get through and meet the boss. Is it wiser to manipulate closer to the 93 step RNG per reset, or don't bother and go for approximately 70 each time? The 93 steps is still the way to go. Just because you finish some leg of the journey doesn't mean that, suddenly, the remaining 69 steps are completely worthless. You were at C33B:5E when you made the P-FROG bandit drop dead. The next encounter was at 7C. The difference between the two numbers, 0x5E and 0x7C is 30 steps. By resetting when you did, you effectively wasted 29 steps. Ouch. Keep that in mind. Just because you beat a boss doesn't mean the leftover RNG is suddenly bad.
You mean this, at the bottom of the post with an edit as an afterthought?
The chosen action might change depending on earlier stuff throughout the battle, but I haven't been able to reset to change that.
Speaking of, why did you set things such that you have three different characters that each must survive at SEI-RYU? One with GAZE and POWER, another with POWER, and the last with TELEPOR.
Somehow, ESP blocks the physical attack of SEI-RYU on the first turn. I've had success with that in my tests against SEI-RYU, if you want to keep that in mind.
As far as I can tell, ESP is not "bugged" after all... it works exactly like a shield. The "effect bytes" used by shields and ESP do nothing except print a message--the defensive functionality comes from the flags byte, not the effect byte. ESP has a different effect byte from shields because it prints a different message.
Since the USA version doesn't allow such memory corruption, and even with it, forgoing use of such a glitch shows so much more of the game, enough to hopefully justify publication alongside.
I just looked at how much mana your character currently has. Here's the projected values at the start of each battle:
I'll look into the encounter rates for a bit. At least have some numbers we can use to plan out where we want the RNG.
We use a starter HAMMER from a MUTANT M to get the extra damage we need. No need to buy any weapons beyond the needed BATTLE sword. Or, for that matter, sell anything. I thought I uploaded this, but I couldn't seem to find any links in my previous posts. Well, here's a new upload.
Again, to force a single enemy against STEWARD's cronies, C353 must be one of these: 1B 2A 2B 2C 35 4E 52 64 6D 7A 8D 94 97 99 9F A1 A2 A3 A6 A7 AF C4 C5 C8 CA E0 FC FD FE
When you have a nice party with the necessities, cheat the random battles away. Unless you really want to experiment with avoiding battles, the point of the test run is to see how we can deal with the battles we must fight. Naturally, you won't be able to record the cheating all that well in a movie, but being able to stroll straight to bosses without intervening battles speed things up greatly.
Then POWER: C30B:95 (POWER, BARRIER, nil, HP+) C30B:96 (Def+3, POWER, BARRIER, nil) C30B:97 (nil, Def+2, ESP, POWER)
At your Mana level, it shouldn't matter how much Mana you gain from STEWARD's cronies, even zero is fine as long as you guarantee +5s from later battles.
Oh, wait. You don't seem to recall this is a requirement in my plans. It won't work without a second POWER user, so yes, we absolutely need one!
All that's happening with the swamps is the fact the encounter rate changed. There's 14 steps of swamp to go through. Just set the RNG so you have a 14-step gap timed for this encounter rate 13 zone.
I'm finding it difficult to get myself back into gear for this run. Sorry if my posts aren't so full of content like usual. Keep talking, hopefully it'll help me get myself fully aligned again.
The bulk of my plans are here. More RNG details. Although, these were before the discovery of the agility oddity.
The mana bonuses in the TELEPOR route are almost immaterial. At this point in planning, they give practically zero benefit to A. The only real benefit is that we can get away with is smaller Mana gains later.
Also, you messed up starting from the very first battle. You don't fight a LIZARD -- They have 40 HP. A GOBLIN has 20 HP, and is oh, so much more capable of dieing in the first round. If C343 was 0x58 from your title screen reset, possible but difficult, a GOBLIN would have showed up both in the plains and the forest. Again, I'm thinking we walk a few steps first, then reset. We have a few spare steps we can pick to try to mess with the RNG here.
local stepstoreset = math.floor((frame-startframe-20)/16)
local stepsafterreset = 185-stepstobattle
local totalsteps = stepstoreset + stepsafterreset
local wastedframes = frame-startframe-20 - 16*stepstoresetExpect a new run started in a day or two.
I would greatly prefer being able to at least somewhat identify what's happening. Short names, please.
C30B|Ab1|Ab2|Ab3|Ab4
----+---+---+---+---
86 | - | + | + | -
93 | - | + | - | +
166| - | /2| + | /2
243| - | + | - | +
-: Subtracts initial number of uses (bottoms out at 0).
+: Adds initial number of uses.
/2: Divides number of uses by 2, rounding down (adds 1 use if no uses left).
=: Restores number of uses to initial value (not used).C30E: | | 87 | 55 | 39 | 36
|Original+---+------+---+---+------+---+---+------+---+---+------+---
| uses |new|action|RNG|new|action|RNG|new|action|RNG|new|action|RNG
-------+--------+---+------+---+---+------+---+---+------+---+---+------+---
Abil. 1| 10 | 20| + | 53| 5 | /2 | 10| 10| = |213| 0 | - |212
-------+--------+---+------+---+---+------+---+---+------+---+---+------+---
Abil. 2| 3 | 6 | + | 76| 6 | + |115| 0 | - |211| 3 | = |222
-------+--------+---+------+---+---+------+---+---+------+---+---+------+---
Abil. 3| 10 | 0 | - |182| 10| = |246| 10| = |233| 20| + |112
-------+--------+---+------+---+---+------+---+---+------+---+---+------+---
Abil. 4| 20 | 10| /2 | 40| 0 | - |151| 10| /2 | 8 | 20| = |213C30B| +1| +2| +3| +4| +5
----+---+---+---+---+---
86| 53| 76|182| 40| 66
----+---+---+---+---+---
93| 64|197|124|166|216
----+---+---+---+---+---
166| 42| 44| 22|247|162
----+---+---+---+---+---
243| 75|193| 80|138| 12C30B|+1|+2|+3|+4|+5
----+--+--+--+--+--
86| +| +| -|/2| +
----+--+--+--+--+--
93| +| -| +| -| =
----+--+--+--+--+--
166|/2| +|/2| =| -
----+--+--+--+--+--
243| +| -| +| -|/2C30B|Ab1|Ab2|Ab3|Ab4
----+---+---+---+---
86 | - | + | + | -
93 | - | + | - | +
166| - | /2| + | /2
243| - | + | - | +I'll make it explicit -- HUMAN M has 54 Strength. We need 75 to kill in the first shot using a SAW. Only way for HUMAN M to build it is to buy STRONG to permanently get the missing strength, or get equipment with bonus strength attached. STRONG has no effect on anyone who isn't HUMAN M or HUMAN F, so we can't build up a MUTANT M using STRONG.
As far as I know, the "greater chance" is the difference between having it happen and a zero percent chance. Yes, without STEALTH, you can't strike first. I may be wrong here, though.
Also, starting as a mutant gives us this ability right from the start.
Said "unusual situation" would be that we are caught in a battle that can't be done in one turn, we lack any decent (read: still living) humanoid shields, and that 10 defense lets us survive with our paltry HP. I fear this situation won't ever come up.
Although, I'm moving away from *all* direct-damage spells. Roughly 26 Mana for SEI-RYU and 36 for Monster Trio won't really cut it.
An optimal reset takes a shorter time than an optimal run away. Hence, it is almost always worth a reset (and manipulating the "steps to next encounter" RNG right) over attempting to run away.
I'll post the full details later. This post is getting a bit bloated already. Since you are studying the RNG, peek into that for strength bonuses, alright? Assume a two-mutant or three-mutant party, we need at least +6 in two battles without a reset, preferably without switching party order in between. Even better if the bonuses show up in the last slot. In addition, report step-count until next battle.
Is "Add" really so rare?
Those are my battle plans, thus far. Any questions?
column | 0| 1| 2| 3| 4| 5| 6| 7| 8| 9|10|11|12|13|14|15
-------+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--
number | 3| 6| 7|13|11| 1|15|10|12| 9| 8| 2| 4|14| 0| 5
-------+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--
row |10|11| 1|12| 5| 9| 6| 3|15|11| 2| 7| 2| 0|15|15
-------+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--
stagger| 5|15|13| 9| 5| 1|13|11| 1| 3| 3| 7|11|15| 3| 9 - - - - - - -
0 60
- - - - - - -
1 156
- - - - - - -
2 252 252 252
- - - - - - -
3 163 163 163 163
- - - - - - -
4 67 67 67 67
- - - - - - -
5 28
- - - - - - -
6 124
- - - - - - -
7 220 220
- ---> - ---> - ---> - ---> - ---> - ---> -
8 195 195 195 195
- - - - - - -
9 99 99 99 99
- - - - - - -
10 3 3 3 3 3 3
- - - - - - -
11 92
- - - - - - -
12 188
- - - - - - -
13 227 227 227 227
- - - - - - -
14 131 131 131 131
- - - - - - -
15 35 35 35 35 35
- - - - - - -
blank column | 3 belongs | stagger is 5, | continue until | put 252 8 rows | put 220 five | continue until
in column 0, so go down five eight numbers away from 3 rows below the column is
row 10 rows and put are placed filled
35 theren={[0]=3,6,7,13,11,1,15,10,12,9,8,2,4,14,0,5}
r={[0]=10,11,1,12,5,9,6,3,15,11,2,7,2,0,15,15}
s={[0]=5,15,13,9,5,1,13,11,1,3,3,7,11,15,3,9}
RNGTable={}
for c=0,15 do
for i=0,7 do
currrow=(r[c]+i*s[c])%16
currnum=n[c]+32*i
RNGTable[16*currrow+c]=currnum
end
for j=0,7 do
currrow=(r[c]+8+j*s[c])%16
currnum=255-n[c]-32*j
RNGTable[16*currrow+c]=currnum
end
end
---------------------------------------------
--The above part is the good stuff. The following loop just tests that it works.
while true do
for i=0,15 do
for j=0,15 do
currval=RNGTable[16*i+j]
gui.text(20*j,9*i,currval)
end
end
vba.frameadvance()
endThose are my battle plans, thus far. Any questions?