Pocky & Rocky / Kiki Kaikai: Nazo no Kuro Manto

I fixed a small thing that can possibly lead to a bug in sd2snes:

I'm not fluent enough in ASM to even notice the change you made, but I'll be sure to look into it later after I've done more debugging of my code


I've managed to make it work properly all the way up to the first boss transition now. there's some funny graphical glitches on the way there that I need to sort out too... the lives/bombs/score etc HUD is scrolling around and broken.


The first boss music is called by a different address so it's likely there's some error relating to that.
it seems to be RTL to the wrong address after running the code.

edit: the graphical glitches in stage 1 are also likely a product of this. i need to get to the bottom of it

Posting what I got before I go to bed anyways.  take a look at my bad code some more

Oh wow, your code is really complex
If you are stuck in a glitch you can send me a geiger savestate for debug

its only complex because i'm not good enough at ASM to make it simple yet.
i would really like to learn how to better translate the databanks to track numbers..

the thing i got stuck on was that the same tracks appear repeatedly throughout the game and so should all be the same number in the end.

the flow of that code is "determine which databank is loaded, then what track, then assign number" but it came out a giant mess lol. it does work so i shouldnt complain so hard

Alright. We're ready for a beta release.


Known issues: 
Title screen music does not end when entering demo play. 
Fadeout on music after boss not present. 
Track dimming on pause not working.
Songs are too loud compared to original music
I think the map screen is too short? it seems like it ends much faster than an unpatched rom.

Also, I didn't bother to make proper loops out of any of the audio or even edit it in any way.

I've attached the patch, assembly, etc to this post. 

I think you can get the music pack from here, for now

https://www.dropbox.com/s/mwznwh5kkdauwzd/par_msu-music.zip?dl=0


And one last thing: 
Stage Select code is on the character select screen, 
hold Y and X. 
press  A A A A B B B B A B A B A B A B. A jingle should play. Now select your character.  This can come in handy while testing.

I took a glimpse at your asm to check for errors (it is too complex though to check everything).

-----------------------
1)
LDA $2000; tell msu1 to play?
AND #$08
BEQ $01
jmp spcfallback

This checks for the missing track, if it is not missing it will branch 1 bytes (in my original asm it was RTL). Since you jmp elsewhere you must use BEQ $03 (bytes)
(jmp = 3 bytes- same bank; jmp.l= 4 bytes- long jump to different bank )
------------------------
2)
You have much redundant code since bsninky makes it different:

This is my code to check the audio busy.
loop:
Bit $2000 ;checking to see if the MSU1 is... ready
BVS loop

This is the same like:
pha
lda $2000
BIT #%01000000 ; check Audio Busy (bit $40
BNE spcfallback ; wait for audio busy flag to turn off?
pla
Your code won't work on sd2snes, since it makes always spc fallback here. On emulators you have the audio immediately, therefore this works in your tests.

---------------------------------
My " previous" code
AND #$08
BEQ $01
RTL
(needs now to be corrected to BEQ $03 jmp spcfallback)
does the same like
BIT #%00001000 ; check for "missing track" bit,
beq trackfound
stz $2007 ;stop MSU
bra nomsu
Here you check bit $08 for track missing (that what I did above with and #$08. STZ $2007 (stop current play) is a good thing, can be added to my code (it is already induced with sta $2004, stz $2005 I think).

----------------------------------
This code has has mistakes btw:
pha
lda $2000
BIT #%01000000 ; check Audio Busy bit
BNE spcfallback ; wait for audio busy flag to turn off?
       pla ;you load the stack into A
BIT #%00001000 ; and check this now for "missing track" bit,
beq trackfound
stz $2007 ;stop MSU
bra nomsu
----------------
correct would be
pla
LDA $2000 (load msu status again)
BIT #%00001000 ; check for "missing track" bit,

But your code for spc fallback doesn't make much sense at all, since you loop always to spcfallback. This actually is the msubusy loop, and not the spcfallback:
spcfallback:  ; i stole this bit from bsinky. this should be spc fallback
pha
lda $2000
BIT #%01000000 ; check Audio Busy bit
BNE spcfallback ; wait for audio busy flag to turn off?
pla

SPCfallback means you unmute the spc. By the way, this will horribly crash on sd2snes. you push each BNE spcfallback to stack without pulling it (better renounce on stack operations here).

To implement spc fallback you need to store the correct pointer to $10, $11, $12 again, run the audio routine and thereafter pick the correct theme. I would not implement spc fallback for this game and rather have a complete pcm pack.

----------------------------------------------

the thing i got stuck on was that the same tracks appear repeatedly throughout the game and so should all be the same number in the end.

If some pcm are the same but have a different index, like track 4 is the same but have different pcm indexes alike $04, $14, $24... you can before storing to 2004:
pha
AND #$0F ; isolate lower 4 bit
cmp #$04 ; is it 04?
BNE continue
PLA ; clear stack
LDA #$04 ; load track 4 for all indexes with lower byte 4
BRA $01 ; branch over load index from stack (we cleared the stack above)
continue:
PLA ; load the track index in case it is not track 4 requested
STA $2004
STZ $2005

Conn wrote:loop:
Bit $2000 ;checking to see if the MSU1 is... ready
BVS loop

That's a lot better than what I came up with to check if MSU-1 audio was busy, I'll have to remember that one!

I haven't looked at the full ASM, but fair warning if you used any of my attempted SPC fallback code: it's pretty buggy. For the Shin Megami Tensei MSU-1 patch, SPC fallback works sometimes, and other times isn't able to mute SPC audio when the next MSU-1 track starts, so you end with overlap (or some bug like that, I haven't tested that in a while.)

It's great to see progress being made on this one, good luck with the rest of the way!

Thanks for the in depth breakdown of what that code actually does, Conn.

My absolute beginner level makes it hard to figure out what is actually being done. I haven't ever read asm before this project and I also haven't done any programming in... 20 years.

I'll look at your suggestions and try to clean things up today. I'm just impresssed that my rats nest of code actually worked.

however, even with my bad code, i did accomplish spc fallback working in emulator. i did this by not changing the databank being loaded. 0 on any of the databanks in this game is a mute track, so you can write back the prior accumulator data to 2140 nd the correct track plays.

bsinky wrote:SPC fallback works sometimes, and other times isn't able to mute SPC audio when the next MSU-1 track starts, so you end with overlap (or some bug like that, I haven't tested that in a while.)

in my order of operations, 

I mute SPC, Then figure out what track to play, Then try to play via MSU, then I start SPC back up if MSU fails. This might help you sort out your situation.


i've tidied up this section thanks to your help Conn. it looks much nicer now, thank you.

Code:

;-----msu code stuff

LDA $0176 ; loads track to be played.
 STA $2004 ; writes track number to MSU-1
 STZ $2005 ; writes a zero to high byte? so it looks like 00XX where XX is the track number
 STZ $2006 ; zero to 2006. no volume. stops static/noise from msu during track loading i think?

loop:
Bit $2000 ;checking MSU1 busy
BVS loop
LDA $2000;
AND #$08 ; is the track found?
BEQ $03 ;if yes, don't play spc.
jmp: playspc

;---- if you have no loop tracks
;lda $0176
;cmp #$xx ;1st noloop track
;beq noloop
;cmp #$yy ;2nd noloop track
;beq noloop
;.......
LDA #$03 ;all loop
bra $02
noloop:
LDA #$01
STA $2007 ; sets MSU play
LDA #$FF ;\raises volume to maximum
STA $2006;/
rts                  ;rts takes us back to the original SPC muting code, reloads the old values and returns to normal code.

spcfallback:
 stz $2007 ;stop MSU
 lda $170 ; load track number
 sta $2140 ; play spc music
 rts  

I'm trying to clear up the track indexing now, back to my original plan of highbit = databank, lowbit = track

I'm running into a mental problem trying to figure out how to make some not the same with your example code though. for instance, 2 is always different, not every 1 plays the same track, and so on.. any advice?

for some more information:
databanks 1-6 have the same track of 1 so, 11 21 31 41 51 61 are all the same track, but 71 81, etc are not.

for track 3: 23 43 53 63 are the same, other track 3 are differnt
I've been trying to turn this over in my head and I don't understand how to do this with ASM yet.

here's my track map, with sound test numbers on the left, my new schema on the right.

Code:

1  = 12
2  = 22
3  = 32
4  = 42
5  = 52
6  = 62
7  = 13
8  = 24,34,44,54,64
9  = 25,35,45,55
10 = 23,33,43,53,63
11 = 71
12 = 72
13 = 74
14 = 75,82
15 = 81
16 = 16,26,46,56,66
17 = 83
18 = 84
19 = 73,92
20 = 91
21 = a1
22 = 36
23 = 65
24 = b1


I'm on mobile, so writing code isn't easy right now.
With having zero on mute is perfect! You can can then easily have spcfallback

We luckily live in an age now where disk space isn't much a problem, so even if you go for redundant tracks, it doesn't really matter.

I'm impressed what you came up with, with no asm experience.

You should go into AND logics
So my above example:
They you have a track $34 in bit this is
0011 0100; track $34 (high 3 low 4)
Now you want to isolate the low part and make
0000 1111; AND $0f
That results in
0000 0100; = #$04

For isolating the high part you make and #$f0. Just think in bits at this place and logic operaters.

----
Your problem is of course complex.
I'd go for a table here... I can code it for you.

It is like
Pha
PHX
Tax
Lda $20a000,X
Sta $2004
Plx
Pla
Stz $2005

So you have a table at 20a000 of 256 byte length
You transfer the track index from A to X register (tax)  and get the track number from there (lda bank+value in X).
So e.g., track $11 and $31 play track $01 but $f1 plays track $07:

Org $20a000
Db $00, $00,....,$00; bytes 00-0f
Db $00, $01,....,$00; bytes 10-1f
Db $00, $00,....,$00; bytes 20-2f
Db $00, $01,....,$00; bytes 30-3f
...
Db $00, $07,....,$00; bytes f0-ff

Thank you! I think that'll be the last piece in my puzzle of learning how to make these.  It'll be a long time before I understand how that code works though. Expect an updated beta later

I'm so glad I asked for help, you have been invaluable.


Edit: Here's the updated beta! my code looks a lot cleaner now... but no other improvements from last patch.

the map screen seems really short because it looks like its normal behaviour is to detect the end of teh SPC track playing and then move on... i may need to mute that track by setting SPC volume to 0 or something.

music tracks again:
https://www.dropbox.com/s/mwznwh5kkdauwzd/par_msu-music.zip?dl=0

The code looks good, but I think I made some mistakes, sorry

1)
org $20a000 looks a bit late, better have it a bit more compact, like 208500 (doesn't matter in the end however

2)
Where I made a mistake is here:
$20/80D2 8D 04 20    STA $2004  [$02:2004]   A:3202 X:0000 Y:0004 P:envMxdizc

This processor flag tells you that x is on 16 bit read, if it is capital (what we need) it is on 8-bit read. M which is the memory (A) is capital through the Sep #$20.

So what you need to do:

org $208040
PHX; X to stack (native)
php
sep #$20; a in 8 bit X in 16 bit to be on the safe side

LDX #$0000 ; reset X
SEP #$30 ; A and X in 8 bit
sta $0170
lda $10 ;   \Load the databank address into accumulator
sta $0172;   \Write the databank address to ram, as in Databank code
stz $2140; mute audio by setting zero at this time
jsr trackselect  ;track number
PLP
PLX
rep #$20; the PLP already did restored the processor status
lda $0170

REP #$F8
rtl

Then renounce on this here:

LDA $0174 ; loads track to be played.
Pha ;save old values of things
PHX ;/
Tax   ;transfer A to X
Lda $20a000,X ;load from the table, position from X (this table is near the end of the document)
Sta $2004 ;store this to MSU1
STA $176 ; store it to 176 as well so I can watch it for debugging purposes.
Plx ; Restore saved values.
Pla ;/
Stz $2005 ;zero to msu1 high byte. looks like 00XX for track number
STZ $2006 ; zero to 2006. no volume. stops static/noise from msu during track loading i think?

Out of curiousity, how is the status ? If your code is done, feel free topublish it in the database

One small thing I noticed:

Code:


CMP #$32 ; databank 2
      beq data2
CMP #$1e ; bank 3
      beq data3
...
data2:   
   lda #$1f
   jmp sorted
data3:   
   lda #$2f
   jmp sorted
....


this blows unnecessarily up your code, more elegant is something like this:

Code:

   
CMP #$32 ; databank 2
bne $05 ; jump over the next 5 bytes
  LDA #$1f ; (2 bytes)
  jmp sorted ; (3 bytes)

CMP #$1e; databank 3
bne $05
  LDA #$2f
  jmp sorted
....
....


or alternatively if the code is less than 80 bytes you can make an always branch instead of jmp and safe even more bytes:

Code:


CMP #$32 ; databank 2
bne $04 ; jump over the next 4 bytes
  LDA #$1f; (2 bytes)
  BRA sorted; (2 bytes)

CMP #$1e; databank 3
bne $04
  LDA #$2f
  bra sorted
...


The answer is: this isn't ready for a full release just yet. I need to work out a way to pause on the map screen for the song there to end, pause dimming of music, and fade outs for the end of the stage.

It works, and I think it works well, but I've been a bit distracted lately (ffxiv endwalker release)

At any rate, I don't want to call it anything more than a beta until I can correct those three issues that I have found.

Nice, currently I am pretty much busy with AoL, NES (yes, myself086 did it!)
But if you need help and find the time, I'll try to

had a bit of extra time while waiting in queues

I'm having a problem that I'm not sure how to resolve. I think my code looks good for solving the problems i want to solve, but none of it (aside from the hooks) are getting written to the rom, it looks like.

When I inspect the rom in hex vs memory editor in bsnes plus, there's also weird "holes" compared to what I expect to see looking at a hex editor. Is it possible the game is overwriting parts of my code with zeroes? and how would I resolve that?

i'll attach my code to this post but note that i havent been able to debug it yet as i can't seem to run any of it in emulator.

Maybe you have a ips somewhere in your folder.

You also made some stack and other mistakes, e.g.

org $00cbbc
jsl hook6mapmusic ;code to wait for MSU1 track to stop playing before progressing.
;default SPC behaviour is 1(still playing music) or 0 (music play has ended)
nop
nop; you forgot this

I attached a bit corrected asm

Oh. That was absolutely the issue and I just had no idea that was the resulting behaviour. Thanks for clearing it up. and for the corrected code, i'll look at it now and then maybe this gets its 1.0 release.

Pocky and Rocky / Kiki Kaikai Nazo No Kuro Manto


1.0 release:
Tested with a full playthough. Has SPC fallback. Works on unheadered USA rom, may work on others. PCMs just converted straight, no loops created or anything fancy.

Patch:
https://mega.nz/file/PW4SXZxB#d7XvTQAEJAKpzHJBs-kDhzrEk-V9i0NYADIvxX-CxWI

Music:
https://mega.nz/file/HXozEQ7Y#cVRwBZAWFjlF_e_LtM3HTlr6giR1N7UpFypbWIkTI4o

Fantastic
I do not know if many people are going to playtest it, but if you think it is safe enough, feel free to post it straight in the database directly (bugs are also reported there, if any are found - also 2 years later). I am unsure though which genre you want to list it in

oh, i just figured out how to do that! lol.

now it's time to move on to the second project i was working on... something on the "Wishlist." 

stay tuned for a new development thread from me!

I have also big news
*bomb explodes*



J is doing some final playtests, I hope there aren't bad surprises....

oh, cool! i don't know anything at all about converting NES or super gameboy titles yet..

Oh, awesome, thank you for your work Cubear (and Conn too)!
I'll be sure to test the game from A to Z when I'm done with the soundtrack!!

NOTE:
There's no way to add loop points to my audio tracks, because they all have a beginning and an end. Actually, I didn't know you can add loop points? Because, that's not what happened on a real system, the whole track had to loop (I remember playing a lot of PC Engine and MegaDrive CD games)

Also NOTE:
I'll need to reexport my audio tracks with less pause at the end of each track (I think I have 2 seconds by default)