Does anyone have a trackmap for the Chrono Trigger MSU mod? I'm about to start working on a new PCM pack, and can't find one anywhere...

*edit-Found one. Fair warning, it'll be a while before the PCM pack gets publicly released. The composer just finished recording and mastering the set (as in earlier today), and hasn't released a single track yet (I have a copy of the full soundtrack to build the mods off of). They're going to be releasing tracks weekly, and I won't be good to release the mods until the entire album has been released.

Oh, I'm also modding this soundtrack into the Steam remake.

If you're interested in the soundtrack, keep an eye on this Youtube channel:
https://www.youtube.com/c/KaraComparetto

The soundtrack is all acoustic Piano + Pipe Organ + Harpsichord (and sounds AMAZING).

New PCM pack release. 100% acoustic Piano, Pipe Organ, and Harpsichord. This is a truly amazing soundtrack. PCM pack includes a copy of Singing Mountain (track 82), in case you're using a mod that restores that song.

PCM pack: https://mega.nz/file/fZBkiLzD#ypUBSQTa1csf-FQNQkRx9vFxpIhsi2PLARGdG1baLJM



Soundtrack mod for the Steam version of Chrono Trigger:
https://www.nexusmods.com/chronotrigger/mods/22

I i saw she was slowly covering all the chrono trigger. Nice to see she finished and that you got a pcm set up. Let me get my laptop out and i will add it to first post

edale wrote:New PCM pack release. 100% acoustic Piano, Pipe Organ, and Harpsichord. This is a truly amazing soundtrack. PCM pack includes a copy of Singing Mountain (track 82), in case you're using a mod that restores that song.

PCM pack: https://mega.nz/file/fZBkiLzD#ypUBSQTa1csf-FQNQkRx9vFxpIhsi2PLARGdG1baLJM



Soundtrack mod for the Steam version of Chrono Trigger:
https://www.nexusmods.com/chronotrigger/mods/22

What a work, simply bordering on perfection, thank you very much!!!

Sorry for the ignorance, but im new to this, and i want to patch a Chrono trigger Portuguese translated rom with the Orchestral music, But i dont know how to do this :/

Kronicky wrote:Sorry for the ignorance, but im new to this, and i want to patch a Chrono trigger Portuguese translated rom with the Orchestral music, But i dont know how to do this :/

Grab Cthulhu's patch (1.1.4):
https://mega.nz/folder/wIE1gQ5S#rbpeeX5pcNFAXfV9XVt2bA

Take your already translated ROM, and check if it's 4MB or 6MB.

If it's 4MB, you'll use make-HiROM.bat.
If it's 6MB, you'll use make-ExHiROM.bat.

Drag-and-drop your ROM onto the appropriate bat file.

The .bat file will create "chrono_msu.sfc", this is your MSU modded ROM.

Place the MSU modded ROM in the same folder as the PCM set. (you'll also need a "chrono_msu.msu" file in that folder, you can just rename a blank text file this if you don't have one)

edale wrote:

Kronicky wrote:Sorry for the ignorance, but im new to this, and i want to patch a Chrono trigger Portuguese translated rom with the Orchestral music, But i dont know how to do this :/

Grab Cthulhu's patch (1.1.4):
https://mega.nz/folder/wIE1gQ5S#rbpeeX5pcNFAXfV9XVt2bA

Take your already translated ROM, and check if it's 4MB or 6MB.

If it's 4MB, you'll use make-HiROM.bat.
If it's 6MB, you'll use make-ExHiROM.bat.

Drag-and-drop your ROM onto the appropriate bat file.

The .bat file will create "chrono_msu.sfc", this is your MSU modded ROM.

Place the MSU modded ROM in the same folder as the PCM set. (you'll also need a "chrono_msu.msu" file in that folder, you can just rename a blank text file this if you don't have one)

Thank you! That worked! The .BAT file generated two files called "Chrono_msu1" I just had to rename the two files to "chrono_msu" and copy and paste them into the folder where the PCM files were! Thank you very much

Kronicky wrote:Thank you! That worked! The .BAT file generated two files called "Chrono_msu1" I just had to rename the two files to "chrono_msu" and copy and paste them into the folder where the PCM files were! Thank you very much

Glad you got it working!
The name of the files doesn't actually matter, just that all the files; SFC, MSU, and PCMs; have the same file name.

Someone should try to bug qwerty to release his code. It's awesome video player and I almost don't think he ever will complete, so maybe better to ask for a release as it is.

Conn wrote:Someone should try to bug qwerty to release his code. It's awesome video player and I almost don't think he ever will complete, so maybe better to ask for a release as it is.

I have tried a few times. I have mentioned that his 9 video cutscenes are awesome as is and people would love it. Chrono Trigger has 13 different endings, meaning that last video cutscene would be a nightmare to put in. I think we should just skip the one.

Conn wrote:Someone should try to bug qwerty to release his code. It's awesome video player and I almost don't think he ever will complete, so maybe better to ask for a release as it is.

I've his code from 2019 that he shared with me in private in order for me to fix some issues with SPC muting and fading. Some pieces of my code are direct adaptation from his (e.g. Patch.SPCMute and Patch.SPCEcho).

I've not shared/leaked this with anyone out of respect for him; if you can get his explicit permission, I can post it here.
However, it's important to note that his approach to hooking FMVs in consists of adding an unused event to the game scripts via Temporal Flux, and this effectively makes the hack unportable because of how Temporal Flux relocates data (every hack needs to manually implement those events).

I still think that if you're going to go through all this trouble for 2/3 seconds pointless non-interactive videos, it's simply better to have the emulator suspend emulation and play the high-quality videos and audios from the Steam release through something like ffmpeg, which doesn't require converting the video to low-quality tiles and SPC audio data (not to mention synching them both).
You can always fallback to low-quality MSU-1 "playback" for platforms not supporting it (mainly the very niche use of MSU-1 in the original hardware).

Thanks for coming in this discussion Cthulhu.
It's a relief to know you have the asm code. I only have the full video MSU File, which plays the 9 cutscenes.

I messaged qwerty and hope he responds. Question to clear is whether there's a chance that he keeps working on his code or drop it.
Cthulhu you're also professional in asm coding, you know and have qwerty's code... Would it be an option that you fix the issues and progress qwerty's code to a release state? Of course only if qwerty agrees. I'm more or less out in coding due to retirement and don't know CT at all. Never played it

Conn wrote:Cthulhu you're also professional in asm coding

While I often do x86 reverse engineering and somewhat know my way around ARM, this is one of only two projects where I've worked with 65C816. I was often consulting its manual while writing the MSU-1 hack and I am far from an expert (to be honest, very few people today could call themselves a 65C816 expert as most go through it with emulators).

Conn wrote:Would it be an option that you fix the issues and progress qwerty's code to a release state? Of course only if qwerty agrees.

I occasionally check out this thread unlogged, which was pretty much dead until recently, just in case someone runs into a nasty issue with my hack, but I am pretty much done with Chrono Trigger and haven't even played it since 2021.

Haven't run into any issues since and no one has reported anything new, so in regards to the BGM portion of MSU-1, my hack should've completed DarkShock's and qwertymodo's work.

As far as I am aware, he was not satisfied with the quality of the FMVs, so this isn't an ASM issue and neither is the lack of the ending FMV (which is done by editing the game's scripts via Temporal Flux).
His hooking for FMV "playback" should be complete too, albeit, some of his BGM playback won't contain the improvements and fixes from my hack.

As you can see, there is not much for me to do from an ASM side and I can't backport his FMV functionality into my hack without breaking portability, which was the main point for me to write my own hack.

Still, for anyone that cares, I still continue to put emphasis on a non MSU-1 solution, such as ffmpeg.
His video code is around 300 lines, but most of it are reading data from MSU-1 and transfering that data into VRAM and CGRAM via DMA. With an emulator provided video playback functionality, you would only need the hooking of the unused event and the code disabling the attract intro.

In any case, I simply cannot take any hobby projects right now as I am currently working on a portable C++11 netplay framework mainly focusing on retro emulators.

Yes, I also think I remember qwerty's main problem was the video quality. However, also those 13 endings needed to be implemented both in the video and code.

If you one day have the time and interest (I know you aren't much into video, but I still ask)... What do you think if you take qwerty's video Code (with his permission), adjust it to your asm? Then we only need someone who can provide you with savestates like jud6ment so we find out how they are triggered.

It's just a suggestion since I know that it's many people's heart wish to have this patch with FMV support, and we don't know whether qwerty paused or cancelled completely. It's been ages already since he last worked on it.

Even though there is 13 endings, not all of them trigger a video. So that makes it slightly less work. I don't know how many endings we need to show a video, i would have to research (assuming someone wants to do this, actually finish the fmv).

I have no problem with anybody distributing my video player code, that's totally fine. It's just the un-watermarked video file that I don't want released in its current form. I understand that it's disappointing, but this project has always been really special to me, and I'm just not ok with the quality of that video file, and I don't want it out there like that. I've posted a few times before, but just to say it again, what I need to get that project done properly is somebody with experience in CUDA coding (preferably in C++) that can help me port bisqwit's dithering algorithms to CUDA. If I can get somebody to help with that, I can finish the project. If I can't get somebody to help with that, I can't finish. It's really that simple. In the mean time, if somebody wanted to help out with something simpler, I do still need to track down exactly which endings trigger which cutscenes. I have save states for all of the endings, if somebody wanted to load them up and run through them, and record which ones trigger the FMV's, that would be helpful.

qwertymodo wrote:if somebody wanted to help out with something simpler, I do still need to track down exactly which endings trigger which cutscenes.  I have save states for all of the endings, if somebody wanted to load them up and run through them, and record which ones trigger the FMV's, that would be helpful.

Do you mean check the endings in the PS1 version of the game to see what ending FMVs are where? What emulator are the savestates for? I can probably help you do the testing for that.

Yes, I mean watch through all of the endings and see which ones have FMV's. The save states I have are for PCSX-Reloaded. If I remember correctly, I believe I started from a memcard file I found online that had saves just before Lavos for each of the ending conditions, and I think I went through each one and beat Lavos, then took the save state right after landing the final blow. I might also have a cheat file that reduces Lavos' health to 0 (which took me forever to work out on my own because none of the max damage or 0 health cheats I found online actually worked). I believe the save states were made using the PSX 1.1 release of FFC/CT

Conn wrote:However, also those 13 endings needed to be implemented both in the video and code.

No, it needs the tile data in the msu file, a MSU-1 audio stream file and the "Color Crash" event set via Temporal Flux to a script where the FMV should play.

Conn wrote:If you one day have the time and interest (I know you aren't much into video, but I still ask)... What do you think if you take qwerty's video Code (with his permission), adjust it to your asm? Then we only need someone who can provide you with savestates like jud6ment so we find out how they are triggered.

As I've mentioned, it would break portability. I can't automate this stuff without creating a Temporal Flux plugin, which requires C#, and I don't do C# on my own projects. Plus the build process would require the ROM to be loaded into Temporal Flux and the plugin to be called, making the build process for your average user even more complex.

I also doubt I'll reach any better quality than what he has already, and that wouldn't satisfy me. Pretty much 0% chance of me working on this, but if I was to work on it, I would probably fork snes9x or bsnes/ares and implement that playback interface around ffmpeg.

Just one more thing, his hook for "Color Crash" is universal, you just need to add the events in Temporal Flux and he probably have done them all except for the ending FMV. The only thing missing is the msu file with that data and the audio streams, which I can't help.
I don't know what was used to convert frames into tiles; there is a Python script that decodes the dump from the Steam Release and run ffmpeg to generate PNG frames of them, but that's it, I also don't know what was used to sync the frames to the audio stream.

The best I can tell you from the ASM code is the msu file format for his hack.

C struct abstraction:

Code:

struct msu_file
{
    char    magic[4]; // MSU1
    uint32_t offsets[FMV_COUNT];
    uint16_t frames[FMV_COUNT];
    uint8_t  data[]; // VRAM and CGRAM
};

Cthulhu wrote:
C struct abstraction:

Code:

struct msu_file
{
    char     magic[4]; // MSU1
    uint32_t offsets[FMV_COUNT];
    uint16_t frames[FMV_COUNT];
    uint8_t  data[]; // VRAM and CGRAM
};

Not quite.  It's

Code:

struct msu_file
{
    char     magic[4]; // MSU1
    uint32_t offsets[FMV_COUNT];
    struct fmv
    {
        uint16_t frameCount;
        struct frame
        {
            uint8_t tiles[];
            uint8_t palette[];
        } frames [frameCount];
    } fmvs [FMV_COUNT];
};

(apologies if this isn't valid syntax, it's been awhile since I've written nested structs... but hopefully this gets the point across at least);

qwertymodo wrote:The save states I have are for PCSX-Reloaded.

I'm having trouble locating somewhere to download this emulator that I'd consider safe to download from. Don't suppose you could post a link to the emulator?

The only safe place I can find it atm is at:
https://github.com/iCatButler/pcsxr
But there are no releases, and I have no idea how to compile the program from the source code there.

qwertymodo wrote:Not quite.  It's...

Right, the uint16_t frame count info isn't part of the header (it's read after the seek).

I didn't abstract the actual data because while the data written to CGRAM is always 2048, the one to VRAM is variable (41024, 1920, 8192 and 14464) and I was too lazy to fully go through the code and set up an appropriate struct.

qwertymodo wrote:(apologies if this isn't valid syntax, it's been awhile since I've written nested structs... but hopefully this gets the point across at least);

The syntax is fine, but the semantics for flexible array members is not (you can only have the last member be flexible and there must be at least one named member before).
https://godbolt.org/z/Wj7jG5WPq

qwertymodo wrote:I've posted a few times before, but just to say it again, what I need to get that project done properly is somebody with experience in CUDA coding (preferably in C++) that can help me port bisqwit's dithering algorithms to CUDA. If I can get somebody to help with that, I can finish the project. If I can't get somebody to help with that, I can't finish.

CUDA, being a proprietary technology, makes it hard to find knowledgeable people to help around with niche hobbyist projects like this. Add the fact that this person must
also be familiar with those dithering algorithms and your chances of finding someone is pretty dim.
Out of curiosity, why do you need CUDA for this? Is it because the algorithm is way too slow or because your conversion tools are written for CUDA?

Anyway, I've mentioned it here more than once and I've even mentioned this to yourself in the past; if you want to get this project finished in a state that you find satisfactory, just implement a video playback interface around ffmpeg (reserving a few open bus addresses for I/O) and fallback to your low-quality MSU-1 if the interface isn't found; problem solved, 99% of the people running this will be on emulators anyway.
No matter what algorithm you apply, those MSU-1 FMVs will always be a huge downgrade to the other platforms, and unlike hacks like Super Road Blaster, where the frames themselves are interactive, I absolutely don't understand why people are obsessed over a worst version of arguably bad 3 seconds videos.

Here is a concept of how this video playback I/O interface could be:
[8-bits / I] status flags (video file not found, bad video format, etc)
[8-bits / O] controller ports enabled during playback (flags)
[16-bits / O] buttons enabled during playback, pressing one of these buttons cancel playback (flags)
[16-bits / O] video id, writing to the low part starts playback

edale wrote:But there are no releases, and I have no idea how to compile the program from the source code there.

There is a solution file for Visual Studio here: https://github.com/iCatButler/pcsxr/blob/master/win32/pcsxr.sln

1. Install Visual Studio Community: https://visualstudio.microsoft.com/vs/community
   
2. Load pcsxr.sln into Visual Studio (if it asks to fix the solution/project, say yes)
   
3. In the right pane, right click the solution and click "Build Solution"

You might need to ask it to install C++ packages and the Windows SDK in "Tools -> Get Tools and Features".

Cthulhu wrote:

edale wrote:But there are no releases, and I have no idea how to compile the program from the source code there.

There is a solution file for Visual Studio here: https://github.com/iCatButler/pcsxr/blob/master/win32/pcsxr.sln

1. Install Visual Studio Community: https://visualstudio.microsoft.com/vs/community
   
2. Load pcsxr.sln into Visual Studio (if it asks to fix the solution/project, say yes)
   
3. In the right pane, right click the solution and click "Build Solution"
   

You might need to ask it to install C++ packages and the Windows SDK in "Tools -> Get Tools and Features".

I got the C++ package installed, but when I tried to build, it said it couldn't find the Windows 8.1 SDK.

The only Windows SDK options I have to install are all Windows 10 SDKs, so I retargeted the solution to the Windows 10 SDK, tried to build again and got a whole slew of depreciated commands and unable to open include file errors...

I really have no idea what I'm doing when trying to compile this thing...

edale wrote:I got the C++ package installed, but when I tried to build, it said it couldn't find the Windows 8.1 SDK.

The only Windows SDK options I have to install are all Windows 10 SDKs, so I retargeted the solution to the Windows 10 SDK, tried to build again and got a whole slew of depreciated commands and unable to open include file errors...

I really have no idea what I'm doing when trying to compile this thing...

Compiled just fine for me for both Debug and Release builds on VS 2019 after retargeting it to Windows SDK 10 and toolset to 142.
There were some warnings, but nothing really relevant (Release build only gave me 3 warnings).



32-bits Compiled: https://mega.nz/file/0ckgVQKb#q3x9sFpi1AJc1D3ktx6vq6iQ6ulpOpq0bpsQ1TJSm7E