seibaby

Hacker
 View Profile  See their activity

  • Content count

    149

  • Joined

  • Last visited

  • Days Won

    26

Posts posted by seibaby

  7. Posted · Edited by seibaby · Report reply

    It was inevitable, so here we go. While fixing my broken random encounter patch, I discovered a little bug with Random Number Good: overworld encounters are not using Think's RNG. What should have been a call to the RNG to determine whether an encounter should be triggered, is instead a static number, causing overworld encounters to happen with astounding regularity. I forgot how the vanilla RNG worked; it does in fact return a different number each time, but it's using a tainted version of vanilla's RNG system (since Think's code overwrites part of the random number table vanilla used). It may still be random "enough" for encounter purposes, but it's not what it should be.

    The fix is just one byte long, but I'll let think share an updated version of the patch here for inclusion in the next point release. I'll instead share a set of updated patches for the random encounter mod, with the fix baked in, so that it can be properly tested this time.


    Patches (updated to v4)

     

    Apply directly to your 1.9.

    EDIT: With Nowea's help, I re-jiggered some numbers and also went ahead and applied the encounter mod to the overworld as well. This should have the effect of setting the minimum number of steps before an encounter to 9 in dungeons and 5 on the overworld (excepting high encounter areas like forests and WoR deserts). Obviously, this will also fix the faulty overworld encounter logic, so you should notice a difference in overall encounter rate there (increased, according to Mishrak). Dungeon encounter rates should remain the same.

    • Upvote 3

  11. Posted · Edited by seibaby · Report reply

    That should work. Tthere might be other issues that crop up with the ending fix, since it clearly wasn't tested properly, but if that happens you can just switch ROM again. I don't think it should make any difference which patch you apply first, but I would apply BNW first and the PAL patch after. Note that the IPS changes a whole lot more than what is described in the code, so you're probably best off sticking with the code patch I posted.

    I'm not entirely sure what the issue is that the ending fix is supposed to fix. It disables the "wait for music" event command. My hypthesis is that, due to running the game in 50 Hz, the event queue and music becomes desynchronized, meaning that the "wait for music" event command hangs the game, or something. I've never seen the ending glitch in action. I'm not sure why disabling the command causes the Opera to hang, but who knows with FFVI.

    EDIT: re-reading that thread, I don't think the issue is that the Opera HANGS, so much as that the events don't follow the music because the "wait for music" event command is disabled. Desynchronization between music and cutscenes is to be expected from running the game in 50 Hz anyway, there's really not much that can be done about that except rewriting all the cutscenes to be shorter. :P

  12. Posted · Report reply

    You can remove the ending fix from the patch by commenting out the relevant code, namely lines 45-67.

    Also, seeing how it's easier for me to just rewrite the patch for compatibility with my favored assembler instead of learning how to use WLA-DX, I went ahead and did just that. :P
    You can copy the following into a new textfile and save it as xkas-palfix.asm (for example), then download xkas 0.06 from here and follow the above procedure, except you don't have to search documentation for commandline syntax. Just type "xkas xkas-palfix.asm ff3.smc" (assuming your ROM file is named ff3.smc).

    	;***********************************************************************************
; final fantasy 3 pal fix by d4s in 2005! YAY!
;***********************************************************************************
	hirom                     ; Don't change this
;header                    ; Uncomment if your ROM has a header
!freespaceC0 = $C0FEE8    ; Requires X bytes of free space in bank C0
!freespaceC3 = $C3F6F0    ; Requires x bytes of free space in bank C3
	; Fix menu bg layer 2
org $C31454
  JSR Bg2MainScreenEnable
	org !freespaceC3
Bg2MainScreenEnable:
  LDA #$17
  STA $212C
  JMP $D263
	;Uncomment the following to enable the ending fix (breaks Opera?)
;org $C09A4C
;  dw CheckMusic
	;org $C0B8BF
;  db $80                        ; $C0/B8BA A5 EB       LDA $EB    [$00:00EB]   A:0000 X:0188 Y:0000 P:envMxdiZc
                          ; Overwritten by the above (?)
;org !freespaceC0
;CheckMusic:
;  LDA $213F
;  AND #$10
;  BNE AdvanceOneEvent        ; Dont check music if in PAL mode (will crash otherwise)
;  LDA $EB
;  CMP $2141
;  BEQ AdvanceOneEvent
;  RTS
;AdvanceOneEvent:
;  JMP $B8C2

    • Upvote 3

  15. Posted · Report reply

    Don't worry, that comment just seems to be wrong. The assembler directive is correct; 48 banks times 64 kB = 3072 kB, or 24 Mbit.

    WLA-DX is an assembler. It's a program you use to transform human-readable assembly code like that text file into machine-readable byte code like what's in the ROM. Since you found the program and the documentation, you're well on your way to figuring out what to do. If you found only the source code for WLA-DX, I can understand your confusion. What you want is the compiled program, found here:

    http://www.villehelin.com/wla-win32.html

    Basically what you need to do, is put the WLA-DX exe, your ROM, and the .asm file all in the same folder, open up a command prompt (in explorer, shift-right click in the folder), and type in the command to tell WLA-DX to patch your ROM with the .asm file. I can't guide you step by step, because I've never used WLA-DX, but the command syntax should be explained in the documentation.

    • Upvote 1

  16. Posted · Report reply

    That patch is unfortunately incompatible with BNW as-is. It requires free space in the menu code, which is used by BNW for its esper bank system. It can be solved by opening up the .asm file included in the archive in notepad, and changing the offset at line 31 from F091 to F6F0. Then use WLA-DX to assemble the patch. I don't believe it requires any other modifications to be compatible.

    • Upvote 1

  20. Posted · Edited by seibaby · Report reply

    Version 9 of Cover Knight
    - Disables dog block when covering
    - Adds 50% evasion penalty when covering (removes bonus to attackers hit rate from previous versions)

    Requires 91 + 20 + 9 = 120 bytes of free space. You can set the defines individually in case there's no 120 byte block of free space available.

     

      
    
	;Cover Knight and Smart Cover
;by Seibaby
;v0.9 - Disables Interceptor and halves Evasion when Covering
;v0.7 - Fixes a bug where Knight wouldn't reset Defending stance when Covering
;v0.6 - New Smart Cover logic
;       Some Cover nerfs
;v0.5 - Fixes an issue with Knights not taking targets' statuses into
;       account (Image and Clear, but also Zombie and Petrify)
;v0.4 - Fixes a bug causing Knights to Cover monsters attacking other
;       monsters
;v0.3 - Adds an exception to not trigger Cover if the Knight is Near Fatal
;       and the target is healthy.
;v0.2 - Fixes a bug that caused the wrong statuses (on the target) to be
;       considered for whether or not to disable True Knight.
	;Changes the True Knight effect to trigger with a Stamina / 192 chance even if
;the target isn't in Near Fatal status.
	;New Smart Cover patch, which disables True Knight for all attacks
;originating from a player character, unless that character is uncontrollable,
;in which case it will only disable it if the attack comes from a healing weapon,
;or if the weapon is elemental and the target absorbs/nullifies that element.
	;It also considers a few extra statuses for purposes of disabling True Knight.
;ON TARGET: Death, Petrify, Zombie, Magitek, and Image
;           (in addition to Clear)
;ON KNIGHT: Dark, Magitek, Image, Berserk
;           (in addition to Death, Petrify, Clear, Zombie, Sleep, and Muddled)
	
;xkas 0.06
hirom
;header
	!smartCover = $000000       ;Requires 91 bytes of free space
!halveEvade = $000000       ;Requires 20 bytes of free space
!noDogBlock = $000000       ;Requires 9 bytes of free space
	;A few notes on the changes made to these functions:
	;Entering this function, A is 16-bit and X/Y are 8-bit. Register width is
;never changed and the call following this function doesn't care about Carry,
;so I have removed a bunch of useless PHP/PLP and REP #$20 throughout, to make
;room for the new code.
;I have also removed some code related to monsters using True Knight, which was
;supported in vanilla, but unused (and remains so in BNW). The check for if the
;bodyguard was Controlled was also removed (doubly useless).
	org $C2123A
exit:
	;True Knight and Love Token
org $C2123B
trueKnightAndLoveToken:
        PHX
        LDA $B2
        BIT #$0002          ;Is "No critical and Ignore True Knight" set?
        BNE .exit           ;Exit if so
        LDA $B8             ;intended target(s).  to my knowledge, there's only
                            ;one intended target set if we call this function..
        JMP smartCover
        macro smartCover()
        print "Writing smartCover() to ",pc
        reset bytes
        smartCover:
        BEQ .exit           ;Exit if none
        LDY #$FF    
        STY $F4             ;default to no bodyguards.
        JSR $51F9           ;Y = index of our highest intended target.
                            ;0, 2, 4, or 6 for characters.  8, 10, 12, 14, 16,
                            ;or 18 for monsters.
        STY $F8             ;save target index
        STZ $F2             ;Highest Bodyguard HP So Far = 0.  this makes the
                            ;first eligible bodyguard we check get accepted.
                            ;later ones may replace him/her if they have more
                            ;HP.
        PHX     
        LDX $336C,Y         ;Love Token - which target takes damage for you
        BMI .noLove         ;Branch if none do
        JSR evalKnight_skip ;consider this target as a bodyguard (skip Stamina
                            ;and Near Fatal checks)
        JSR newTarget       ;if it was valid, make it intercept the attack  
.noLove PLX
        LDA $3A36
        BNE .exit           ;Exit if Golem is active
        CPX #$08            ;Check attacker
        BCS .status         ;Branch if attacker is a monster
        CPX $F8
        BEQ .exit           ;Exit if Attacker = Target
        LDA $3EE4,X         ;Attacker status byte 1-2
        BIT #$2002
        BNE .heals          ;Branch if Muddled or Zombied
        LDA $3394,X         ;Check if Attacker is Charmed
        BMI .exit           ;If not Muddled, Zombied, or Charmed, this
                            ;attack was initiated by the player, so exit
.heals  LDA $11A9           ;Special weapon property
        AND #$00FF          ;Isolate bottom byte
        CMP #$0018          ;Check "Curative Attributes"
        BEQ .exit           ;Exit if set
        SEP #$20         
        LDA $11A1           ;Attack element(s)
        PHA
        XBA
        PLA                 ;Copy to high byte
        REP #$20
        AND $3BCC,Y         ;Target absorbed/immune elements
        BNE .exit           ;If any absorbed or nullified, exit
.status LDA $3EE4,Y         ;Target status byte 1-2
        BIT #$04DA
        BNE .exit           ;Branch if Death, Petrify, Clear, Zombie, Magitek,
                            ;or Image
        JMP trueKnightAndLoveToken_seize
	.exit   PLX
        RTS
        print "smartCover: ",bytes," bytes written"
        endmacro
	.seize  LDA $3358,Y         ;$3359 = who is Seizing you
        BPL .exit           ;Branch if target is seized
        LDA #$000F          ;Load all characters as potential bodyguards        
.cover  CPY #$08
        BCC .saveBg         ;Branch if target is character
        TDC                 ;Null all potential bodyguards
.saveBg    STA $F0             ;Save potential bodyguards
        LDA $3018,Y         ;bit representing target
        ORA $3018,X         ;bit representing attacker
        TRB $F0             ;Clear attacker and target from potential
                            ;bodyguards
        LDX #$12
.loop   LDA $3C57,X         ;High byte = Relic Effects 3
        ASL #2              ;Check bit 6 (True Knight)
        BCC .next           ;Branch if no True Knight effect                
        LDA $3018,X
        BIT $F0
        BEQ .next           ;Branch if this candidate isn't on the same
                            ;team as the target
        JSR evalKnight      ;consider them as candidate bodyguard.  if they're
                            ;valid and their HP is >= past valid candidates,
                            ;they become the new frontrunner.
.next   DEX
        DEX
        BPL .loop           ;Do for all characters and monsters
        LDA $F2
        BEQ .exit           ;Exit if no bodyguard found [or if the selfless
                            ;soul has 0 HP, which shouldn't be possible outside
                            ;of bugs].
        JSR newTarget       ;make chosen bodyguard -- provided there was one --
                            ;intercept attack.  if somebody's already been
                            ;slated to intercept it [i.e. due to Love Token],
                            ;the True Knight will sensibly defer to them.
.exit   PLX
        RTS
	
;Make chosen bodyguard intercept attack, provided one hasn't been marked to do
;so already.
	newTarget:
        LDX $F4
        BMI .exit           ;exit if no bodyguard found
        CPY $F8
        BNE .exit           ;exit if $F8 no longer points to the original
                            ;target, which means we've already assigned a
                            ;bodyguard with this function.
        STX $F8             ;save bodyguard's index
        STY $A8             ;save intended target's index
        LSR $A8             ;.. but for the latter, use 0,1,2,etc rather
                            ;than 0,2,4,etc
        LDA $3018,X
        STA $B8             ;save bodyguard as the new target of attack
        SEP #$20
        LDA $3AA1,X
        BIT #$02
        BEQ .noDef          ;Branch if not Defending
        JSR $0A41           ;Clear Defending flag
        JSR $0A3C           ;Relax Defending pose
.noDef  REP #$20
.exit   RTS
	
;Consider candidate bodyguard for True Knight or Love Token
evalKnight:
        LDA #$0020
        BIT $3AA1,X
        BNE .exit           ;Exit if in back row
        LDA $3EE5,Y         ;Low byte = Status byte 2
        LSR #2              ;Check bit 1 (Near Fatal)
        BCS .skip           ;Skip Stamina check if target Near Fatal        
          LDA $3EE5,X         ;Knight's Status byte 2
        LSR #2              ;Check Near Fatal
        BCS .exit           ;If Knight is Near Fatal, exit
        SEP #$20                
        LDA #$C0            ;192
        JSR $4B65           ;Random: 0 to 191
        CMP $3B40,X            ;Stamina
        REP #$20
        BCS .exit           ;Exit if Stamina was lower
.skip                       ;Love Token enters here
        LDA $3AA0,X
        LSR     
        BCC .exit           ;Exit function if entity not present in battle?
        LDA $3358,X         ;$3359 = who is Seizing you
        BPL .exit           ;Exit if you're Seized
        LDA $3EE4,X
        BIT #$B4DB          ;Death, Petrify, Clear, Zombie, Sleep, Muddled,
                            ;Dark, Magitek, Image, Berserk
        BNE .exit           ;Exit if any set
        LDA $3EF8,X     
        BIT #$3210          ;Stop, Freeze, Spell Chant, Hide
        BNE .exit           ;Exit if any set
        LDA $3018,X
        TSB $A6             ;make this potential guard jump in front of the
                            ;target, can accompany others
        LDA $3BF4,X         ;HP of this potential bodyguard
        CMP $F2         
        BCC .exit           ;branch if it's not >= the highest HP of the other
                            ;bodyguards considered so far for this attack.
        STA $F2             ;if it is, save this entity's HP as the highest
                            ;HP so far.
        STX $F4             ;and this entity becomes the new bodyguard.
.exit   RTS
print "Cover function end: ",pc
print "Vanilla Cover ends: c212f4"
	;Check for Covered attacks in Hit Determination
	;Disable Dog Block if attack was Covered
org $C22282
checkDogBlock:
        LDA $3EF9,Y
        ASL
        BPL C22293            ;Branch if not dog block
        JSR skipDogBlock
        macro skipDogBlock()
        print "Writing skipDogBlock() to ",pc
        reset bytes
        skipDogBlock:
        CPY $F4
        BNE .exit
        CLC
        RTS
.exit   JMP $4B53           ;0 or 1          
        print "skipDogBlock: ",bytes," bytes written"
        endmacro
;        BCC C22293          ;50% chance
;        LDA #$40      
;        STA $FE             ;Set dog block animation flag
	org $C22293
C22293:
	;Halve Evasion if attack was Covered
org $C22345
        JSR halveEvasion    ;Get Evade
        macro halveEvasion()
        print "Writing halveEvasion() to ",pc
        reset bytes
        halveEvasion:       ;Y = target of attack
        CPY $F4             ;Index of bodyguard ($FF if no bodyguard)
        BNE .exit           ;If not the same as target, exit
        LDA #$FF            ;255
        SEC
        SBC $3B54,Y         ;255 - (255 - Evade * 2 + 1)
                            ;(= Evade * 2 - 1)
        INC                 ;Evade * 2
        LSR                 ;Evade
        LSR                 ;Evade / 2
        JMP $2861           ;New blockvalue from halved Evade
.exit   LDA $3B54,Y         ;(255 - Evade * 2 + 1)
        RTS
        print "halveEvasion: ",bytes," bytes written"
        endmacro
	org !smartCover
%smartCover()
org !halveEvade
%halveEvasion()
org !noDogBlock
%skipDogBlock()
    • Upvote 3

  24. Posted · Report reply

    I need to update Cover Knight to fix a few things. Dog block shouldn't synergize with Cover, and in order to fix that, I need to poke into the Hit determination code and add some Cover checks there. While I'm at it, I might as well amend that code to properly halve Evasion when Covering rather than doubling attacker Hit Rate, and whatever else we agreed on in the Cover discussion thread.

    I'll have time to work on that next week.

     

Affiliates

Other Resources