Unfinished New Game +

If you are stuck in the Dunan Unification Wars; or wish for more details on the gameplay systems, this is the place.
Post Reply
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

I'm not quite sure myself. I am not very programming-oriented, so I have to go through trial&error. To start easy, I chose to first check the portraits, and the save state gives the right picture while the decompressed binary gives the left. Hm, I guess I'll simply have to spend some more time and really understand this.
Untitled-1.jpg
You do not have the required permissions to view the files attached to this post.
User avatar
Pyriel
Webmaster
Posts: 1227
Joined: Wed Aug 18, 2004 1:20 pm

Re: Trial Road

Post by Pyriel »

What color table did you use for the decompressed one? I have a bunch of stuff like this that I extracted directly from the game, and it came out fine once I found the right CLUT. Another possibility is that you didn't feed it the right array of compressed data. Like maybe you passed in a buffer that still had the compressed size in it? Something like that could cause some compressed block to be recognized properly, while others just come out a mess.
You do not have the required permissions to view the files attached to this post.
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

I went back to RHDN's discussion and finally understand what's up with those "frames"! The decompressor was never meant to be applied to portraits!! DUH! No wonder it comes out all strange. I think I've got it as far as sprites are concerned. Oh beauty!
It would be interesting to adjust the routine for portraits, though. In fact, this will be necessary if the Rune has a face. It shouldn't be too hard seeing how the sprites-routine has already done a pretty good job of it.
Anyway, I'll come back when I have dug out some stuff to show :D
nanami.png
You do not have the required permissions to view the files attached to this post.
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

I am checking through file VJ24 from bottom up, and two sprite sheets looked new and interesting. At least, they show sprites that are not seen in the epilogue dungeon where you can walk around without crashing.
VJ24.jpg
The turtle-like design with eight orbs most certainly is a door to the Rune, and each or each pair would probably correspond to the orb-stand that is situated throughout the dungeon.
EDIT: "turtle-like" is just my impression; they are in fact eight snake-heads, each with an orb in the mouth.
EpilogueDungeon.jpg
I could find only three up to now (marked with a red circle), and I know that two of them were in 09-11 while I found another one when I chose 09-13 as destination (if I remember correctly). Going through 09-14 to 09-17 may lead to more of them (Floors F4 and F5 look suspicious!).
If the palette is correct, there is also a save sphere that is a different color from the usual save spots.
This kind of reminded me of the game "Xenogears", where save spots actually turned out to have a meaning. Maybe in Suikoden also, the save spots' identity might have been low-rank runes of this Rune of Circulating Path (or whatever the name).
Anyway, there were some clouds in addition to all that, so the final location might have been the very top of a tower.
You do not have the required permissions to view the files attached to this post.
User avatar
Raww Le Klueze
Global Admin
Posts: 1915
Joined: Sat Jun 26, 2004 1:38 am

Re: Trial Road

Post by Raww Le Klueze »

The purple Journeyman Crystal could be the sprite representation of this "rune", but I wouldn't draw any conclusions about it's relationship with the blue ones from that as palette swaps are mostly a time saving thing.
Doctorum Non Urina Singulus.
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

You are right about that.
It would be neat, though, since the ordinary "Journeyman Crystal" are also a point where the player changes the future of the game play, which kind of links to the nature of the Rune.

Just an addition about the decompressor: it works on portraits!
Having gone through the routine, one by one (such a cool mechanism, too, it's the first time I've taken a close look at something like this) I realized that the frame size shouldn't matter. It was probably a convenient size to use for character sprite animations, but portraits don't move and so there is really no reason to use a larger frame. So I re-ran the decompressor on the portrait, and the output came out perfect. Puzzled, I went back to the old copy of the project (I had made a new one to work on sprites), and realized that the old project *wasn't producing any output*. Meaning, I was looking at the initial mistake (as Pyriel pointed out, it could be reproduced with the compressed size included) over and over wondering why the routine fails on portrais. :shock: DU-UH-UH!!!
I still have no idea why the old project doesn't work, but I must have messed up the environment somehow. Creating too many "test" projects and a sleepy head does that, I guess...

Well, that cleared, I can go hunt for the Rune's face :D
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

Hm, I went around looking for stuff, and I'm afraid this might be pretty much it.

I checked the SDD and VAB files, consulted the internet (seems like I see your name everywhere, Pyriel!), and found out they are sound files. Why they'd include a 102MB XA file in addition to the sequences, is beyond me. All the compression and decompression - just to make room for Annallee's band, possibly? Well, they must have had their reason.

Anyway, with SDD and VAB files out of the way, there were only BIN files left to look into. I decided to only rip 8bpp sprites, which seemed like a safe route to take, compared binaries, ripped through VJ17 (Sajah), VJ19 (L'Renouille castle outside), VC12 (Sindar ruins) and VC14 (Sindar ruins), but to no avail. 8bpp sprite sheets of VJ18,20-23 looked identical to those included in VJ24 (made visual comparison of bit images), so I didn't bother go through them.
I saw Elza's portrait at the end of VJ17, so if the Rune had potrait data, it should have been in VJ24. Unfortunately, it's not. Another disappointing find is that VJ17 didn't reveal what's inside that hidden cave. If they left a whole dungeon in the game, you'd think they wouldn't care if some additional sprites remained, but no. Of course, it's always possible that they were never made.

Since I don't quite feel like hunting the 210_BOSS folder for a possibly unknown boss (unlikely), I am at a loss for what else could be done :(
I'm glad I could share some finds, though. Two weeks ago, I would never have thought I'd actually be ripping sprites myself, but it was lots of fun learning this stuff despite the tedious work involved :D
User avatar
Pyriel
Webmaster
Posts: 1227
Joined: Wed Aug 18, 2004 1:20 pm

Re: Trial Road

Post by Pyriel »

When I ripped the data for the Bestiary years ago, those files were parsed too. At the time, I wasn't terribly familiar with the game, but nobody ever indicated that we'd dug up an unknown boss. We found the Chimaera in Tenzen, and a couple of other oddities, along with dozens of different Highlands Soldiers, but not much else.

I honestly had a feeling the data was gone, never created, and/or full of placeholders. Given what we know now, it seems odd that any part of the dungeon would be populated with CutRabbits and FurFurs that haven't had their stats bumped much, if at all. So the whole thing was probably scrapped at the point of actually implementing its scenario and integrating the events with the rest of the game. Probably a few people assembled the level, and filled it with placeholder enemies. Then before more appropriate assets could be completed, the place was scrapped for time or because it was too complicated. Given the limitations of the PSX, they might have even scrapped it if what they were planning meant forcing the save to take up another block.
User avatar
Celes
Global Admin
Posts: 1422
Joined: Sun Nov 12, 2006 6:25 pm

Re: Trial Road

Post by Celes »

Did you find any fragments of the actual algorithms which grants the reboot while keeping you stuff?
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

Oh, then I'm glad I didn't go into the BOSS folder! lol

Going into the booting algorithm is way too difficult for me, but your suggestion did get me to some further finds:

I see G1LOAD.BIN which obviously loads the S1 data, and BLOAD.BIN which seems to stand for BOOTLOAD, and my first guess would be to load the usual save data. However though, it is intriguing that I see bunch of "clear_event"s in the beginning of the BLOAD. That's something you'd expect for a newgame+.

Also, I notice that the region $23F2C-$30E3F in VJ24 look like copies of $250FC-$3230F in BOOT.BIN.
G1LOAD also shares some part of this: $1B3F4-$260FF in G1LOAD look like $26134-$30E3F in VJ24.
If I had to hazard a guess, I'd say these are all the flags of this game.
The other event BINs from VA-VK don't seem to share this part (I checked VJ17, 18, 20-23 in particular; other than that, I only picked out random files)
What's further intriguing: BLOAD *does not* share this data, either.
So there is definitely something special and bootable about VJ24! - And maybe, just maybe, BLOAD goes with it.

But these are just observations I make by looking at the bit images of these files, and I simply have no idea how to proceed from there. I'd have to start with Assembly Language 101.

Apart from reading Assembly, things that might still be possible are:
plan A) Disable the event at the top level of the dungeon so it doesn't crash, and see everything in action (hopefully)
plan B) Manually recreate the map after the above-mentioned crash-point, just to get the visuals.
I'd have to find out how a map is laid out in the game, though!

EDIT: I made a quick check of the shared regions of VJ24 and BOOT, and although the structure itself is the identical, contents are different. So these might indeed be flags. Analysing the flags of VJ24 (and maybe comparing it to a save file?) might reveal where the newgame+ starts. My first place to check would be the Sindar Ruins :D
EDIT: One more thing, the region shared by all three of VJ24 and BOOT and G1LOAD are identical, so only $23F2C-$26133 in VJ24 matter.
EDIT: corrected to show address in EN version.
Last edited by ne__ya on Tue Dec 20, 2016 9:56 am, edited 1 time in total.
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

OK, I couldn't drop this stuff so went on investigating.
The comparison of similar regions are summed up below (I hope the table comes out readable):

Code: Select all

				VJ24				BOOT				 G1LOAD				BLOAD
--------------------------------------------------------------------------------
Region A	$23F2C-$26133	 $250FC-$27303 	NA				   NA

Pointers	NA					$27304-$27603	 NA				   NA

Region B	$26134-$30E3F	 $27604-$3230F	 $1B3F4-$260FF	 NA

Region C	NA					NA				   $8A9C-$1B3F3	  $98E8-$1C23F
EDIT: corrected to show address in EN verion
Region A: not a flag region. VJ24 and BOOT have differences, but "A" taken from a mid-game save state had no difference to BOOT.
Region B: All three are identical.
Region C: G1LOAD and BLOAD are almost identical, save for 6 pointers and 2 other places in the binary.

I took a look at memory card data, but the stuff in there didn't seem similar to any of these. I wonder if they are compressed, too? They are such tiny data.
I should have learned by now that things aren't that easy to find!
Antimatzist
Posts: 2770
Joined: Sat Apr 19, 2008 9:48 am
Location: Germany, yeah baby
Contact:

Re: Trial Road

Post by Antimatzist »

fyi, you guys are now famous, Pyriel and ne__ya ;) http://kotaku.com/after-18-years-fans-f ... 1790331169
User avatar
Nikisaur
Posts: 462
Joined: Fri Apr 16, 2010 5:26 am
Location: New Zealand

Re: Trial Road

Post by Nikisaur »

Awesome work guys! This is actually so interesting.

I don't know much about code stuff, but from story/world perspective I'm looking at the "Rune of Circulating Paths". Is anyone able to confirm if it's the same as Hikusaak's Circle Rune? My brain is swelling with theories of multiverse/Hikusaak clones and connection to the Sindar Ruins. I also really like the idea of Journeyman Crystals being connected to the Rune of Circulating Paths (canonical saving and NG+?whaaaaat)

Anyway, good job team, I'm sad this didn't actually exist in the final game though :(
The only thing Suikoden lacks...is dinosaurs.
User avatar
Pyriel
Webmaster
Posts: 1227
Joined: Wed Aug 18, 2004 1:20 pm

Re: Trial Road

Post by Pyriel »

ne__ya wrote:OK, I couldn't drop this stuff so went on investigating.
The comparison of similar regions are summed up below (I hope the table comes out readable):

Code: Select all

				VJ24				BOOT				 G1LOAD				BLOAD
--------------------------------------------------------------------------------
Region A	$23F2C-$26133	 $250FC-$27303 	NA				   NA

Pointers	NA					$27304-$27603	 NA				   NA

Region B	$26134-$30E3F	 $27604-$3230F	 $1B3F4-$260FF	 NA

Region C	NA					NA				   $8A9C-$1B3F3	  $98E8-$1C23F
EDIT: corrected to show address in EN verion
Region A: not a flag region. VJ24 and BOOT have differences, but "A" taken from a mid-game save state had no difference to BOOT.
Region B: All three are identical.
Region C: G1LOAD and BLOAD are almost identical, save for 6 pointers and 2 other places in the binary.

I took a look at memory card data, but the stuff in there didn't seem similar to any of these. I wonder if they are compressed, too? They are such tiny data.
I should have learned by now that things aren't that easy to find!
What you've found is mostly the LBA listing, which is stuffed into several files on the disc.

The PSX stunk at seeking by filename, so games typically built a list of Logical Block Addresses and sizes for files, and wrote it somewhere, so they could seek more directly. In this case, Konami chose to write the list to several somewheres, including the executable (SLUS*, SLES*, SLPM*, whatever), BOOT.BIN, VJ24.BIN, etc., etc. I think it was written to about 20 files in all. The NA patch has a list of them, since resizing the UWASA files requires correcting the listing.

The other data in BLOAD and GLOAD doesn't appear to be compressed. And in a disassembly, the references to the area are pointers to blocks exactly 512 bytes in size, which could be bitmaps 32x32@4BPP. No idea what the color information would be. That's just a guess.

Edit:

Here's where that block of stuff is referenced.

Code: Select all

RAM:80115588                 addiu   $sp, -0x28
RAM:8011558C                 li      $v0, 0x8007549C
RAM:80115594                 move    $a0, $0
RAM:80115598                 li      $a1, 0x6AC
RAM:8011559C                 sw      $s0, 0x28+var_10($sp)
RAM:801155A0                 lui     $s0, 0x8013
RAM:801155A4                 sw      $ra, 0x28+var_4($sp)
RAM:801155A8                 sw      $s2, 0x28+var_8($sp)
RAM:801155AC                 sw      $s1, 0x28+var_C($sp)
RAM:801155B0                 jalr    $v0
RAM:801155B4                 sw      $v0, dword_80129F18
RAM:801155B8                 move    $s2, $v0
RAM:801155BC                 li      $v1, 1
RAM:801155C0                 li      $v0, 2
RAM:801155C4                 sb      $v0, 1($s2)
RAM:801155C8                 li      $v0, 5
RAM:801155CC                 sb      $v0, 0x17($s2)
RAM:801155D0                 li      $v0, 0x21D0
RAM:801155D4                 sw      $v0, 0x20($s2)
RAM:801155D8                 lui     $v0, 0x8011
RAM:801155DC                 sb      $v1, 3($s2)
RAM:801155E0                 sb      $v1, 0x18($s2)
RAM:801155E4                 lw      $v1, 0x20($s2)
RAM:801155E8                 la      $v0, off_80117508
RAM:801155EC                 sb      $0, 4($s2)
RAM:801155F0                 sb      $0, 5($s2)
RAM:801155F4                 sb      $0, 6($s2)
RAM:801155F8                 sb      $0, 0($s2)
RAM:801155FC                 sb      $0, 0xE($s2)
RAM:80115600                 sb      $0, 8($s2)
RAM:80115604                 sb      $0, 9($s2)
RAM:80115608                 sb      $0, 0xA($s2)
RAM:8011560C                 sb      $0, 0xB($s2)
RAM:80115610                 sb      $0, 0x10($s2)
RAM:80115614                 sb      $0, 0x29($s2)
RAM:80115618                 sb      $0, 0x11($s2)
RAM:8011561C                 sb      $0, 0x12($s2)
RAM:80115620                 sb      $0, 0xC($s2)
RAM:80115624                 sb      $0, 0x15($s2)
RAM:80115628                 sb      $0, 0x16($s2)
RAM:8011562C                 sw      $v0, 0x66C($s2)
RAM:80115630                 andi    $v0, $v1, 0x7F
RAM:80115634                 beqz    $v0, loc_80115644
RAM:80115638                 subu    $v0, $v1, $v0
RAM:8011563C                 addiu   $v0, 0x80
RAM:80115640                 sw      $v0, 0x20($s2)
RAM:80115644
RAM:80115644 loc_80115644:                            # CODE XREF: sub_80115588+ACj
RAM:80115644                 li      $v0, 0x800D4CBC
RAM:8011564C                 lui     $a0, 0x8011
RAM:80115650                 lw      $a1, 0x20($s2)
RAM:80115654                 la      $a0, aSaveDataSizeD  # "save data size %d\n"
RAM:80115658                 jalr    $v0
RAM:8011565C                 sw      $v0, -0x60E8($s0)
RAM:80115660                 move    $a0, $0
RAM:80115664                 move    $v1, $s2
RAM:80115668                 li      $v0, 0x280
RAM:8011566C                 sw      $v0, 0x1C($s2)
RAM:80115670                 sw      $0, 0x614($s2)
RAM:80115674                 sb      $0, 0x628($s2)
RAM:80115678                 sw      $0, 0x62C($s2)
I'm guessing those are common icons or something for the save/load screen that G1LOAD uses when you pull up an S1 save. I think BLOAD is the module for saving and loading S2 data, but I'm a bit foggy on that. I know with 100% certainty G1LOAD is for importing S1 saves.

Files with an LBA listing below. (The startsector property is something the patch process uses later, so it can do raw writes into files that have already been written to the patched image. The fact that it's zero here doesn't mean anything.)

Code: Select all

		  LBAPatchList= {
			 { name="SLUS_009.58",     path="/",     startsector=0,   seekpos=0x7dbbc },
			 { name="VJ24.BIN",     path="/CDROM/100_ARJ/",     startsector=0,   seekpos=0x23f24 },
			 { name="FURO.BIN",     path="/CDROM/140_HONP/",     startsector=0,   seekpos=0x16de0 },
			 { name="HSOUND.BIN",     path="/CDROM/140_HONP/",     startsector=0,   seekpos=0x3cd8 },
			 { name="RBATTLE.BIN",     path="/CDROM/140_HONP/",     startsector=0,   seekpos=0x430a4 },
			 { name="SNDTEST.BIN",     path="/CDROM/140_HONP/",     startsector=0,   seekpos=0x280a4 },
			 { name="IPRG_A3.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x11504 },
			 { name="IPRG_A4.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x11f54 },
			 { name="IPRG_B2.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x1088c },
			 { name="IPRG_D.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x10e20 },
			 { name="IPRG_D2.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x10e44 },
			 { name="IPRG_D3.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x11364 },
			 { name="IPRG_J.BIN",     path="/CDROM/260_IKKI/",     startsector=0,   seekpos=0x10ce4 },
			 { name="BOOT.BIN",     path="/CDROM/270_BOOT/",     startsector=0,   seekpos=0x250f4 },
			 { name="ENDING.BIN",     path="/CDROM/270_BOOT/",     startsector=0,   seekpos=0x3c40 },
			 { name="OVER.BIN",     path="/CDROM/270_BOOT/",     startsector=0,   seekpos=0xbb2c },
			 { name="STAFF.BIN",     path="/CDROM/270_BOOT/",     startsector=0,   seekpos=0x3058 }
		  }, -- end LBAPatchList
Edit 2: For what it's worth, the LBA listing being in a file is about half indicative of something, but then not really. I know that's pretty nebulous, but things like the staff roll from the ending (STAFF.BIN) and the (scroll I think?) shop (FURO.BIN) aren't remotely "bootable", but they have a listing. It's probably just an artifact of development. Of course, it could imply that one of the custom built functions they inserted into a script, e.g., the one in VBsomething.BIN that halves Riou's HP, does some custom loading of files that can't be handled by the existing script commands.

And all that stuff about "events" in BLOAD is almost certainly referring to BIOS events.

Edit 3: Yup. Every reference to those strings immediately precedes a call to printf, and immediately follows a call to the BIOS TestEvent() function. They're just debug messages tracking the system's progress and polling for whether or not an error has occurred.
ne__ya
Posts: 26
Joined: Wed Dec 07, 2016 8:47 am

Re: Trial Road

Post by ne__ya »

Oh. That kind of clears it up, then! I'll add that scrolling further down on G1LOAD also showed the "clear_event" stuff, so as you point out, these are not referring to any game events. Them being BIOS events makes a lot of sense. Thank you always for sharing your insights so generously, Pyriel :D
BTW, do you have any idea what's in the "B" region? This strange structure is very puzzling.
Anyway, looks like it will require digging into the executables and such to find anything meaningful.
(Knowing now that memory card saves only take up $2200 of space, I see how hazardous my guess was. lol)

Contemplating flags raised some additional points:
1) If the newgame+ really starts at White Deer Inn, I wonder if all the prior recruits will be lost, too. Like, the ones you took to the Sindar Ruins? Not that it would matter greatly; you'd go to Muse, then to Coronet and meet Eilie, Rina, and Bolgan. It might come across a bit weird, though.
2) On the other hand, since you'd have the Shield Rune already, there is no need to worry about earlier flags that kept a record of what dialogue you chose in certain places (like when you mark the rock in the very beginning; it replays in Toto when you get the Shield Rune). So you could overwrite all flags without heed. I don't know if that is advantageous at all, but just an observation.

Since it looks like this thread is getting attention, hopefully someone will turn up to help out with the more deep stuff.

>Nikisaur
The "Rune of Circulating Paths" is completely different from Hikusaak's Rune. It's called "めぐる道の紋章" (Meguru Michi no Monshou) in Japanese. Michi is anything like road, way, path. Meguru is a bit more difficult, but means something like going around with circularity and repetitiveness implied. It can be used with things like the four seasons that come and go, your own thoughts that wander around, or even sightseeing going from one place to the next. Monshou is just Rune, "no" means "of".
Post Reply