MM 3D - Model Format discussion

[Twili]

https://mega.co.nz/#!mlIXiLpY!LLaay7YPWFYJmVZOs0jR48dniuyidkSvr7-EqGAGXw4

 

Here's the read-only filesystem of Majora's Mask 3D. The model format is still .cmb, but changes have been made, evidently, as they don't load in N3DSCmbViewer. Not much else to say...

[PwnzLPs]

Awesome! Thanks, Twili!

[PwnzLPs]

Im curious, are the files themselves decrypted? If not, maybe they changed the type of encryption on the files and that's why we can't open them

[Twili]

These are decrypted.

[PwnzLPs]

Ahh, okay. Hmm, has anyone opened them up in a hex editor or anything else?

Edited February 12, 2015 by PwnzLPs

[PwnzLPs]

Sorry for double posting.

 

Most of the CTXB files can be extracted with Texture Ripper 1.1.7, although a whole folder with around 110 or so CTXB files is the exception. The scene folder's CTXB files cannot be ripped using Texture Ripper 1.1.7, so if anyone else can get some information on this, that'd be awesome! I can say this much: What Texture Ripper does is makes completely blank text files for specifically those ctxbs, and doesn't rip anything from them, even if I attempt to do each file individually.

[xdaniel]

WIP:

 

 

Edit, more recent WIP:

 

 

Edit, even better: https://github.com/xdanieldzd/N3DSCmbViewer - no proper release build just yet, tho

[xdaniel]

Not very well tested with other models yet:

 

 

 

[Twili]

I wrote a tool to convert Majora's Mask 3D models to Ocarina of Time 3D models: http://www.mediafire.com/download/je1vzk9e36zpx61/mm2oot.exe

 

And here's the MM3D files all extracted and decompressed: https://mega.co.nz/#!88NWmYJA!IZ9i_k4D14qO4KZy0XzPVm3_oJNtQVcaxxAnA14AmxM

 

Usage: mm2oot.exe model_in model_out

 

Example .bat file to run it (goes in the folder with it):

 

@echo off

mm2oot.exe E:\Users\nick\MyStuff\3ds\extracted_mm3d_rom\actors\zelda2_link_boy_new.gar.lzs\link_demon.cmb E:\Users\nick\MyStuff\3ds\converted\link_demon.cmb

 

 

 

xdaniel is going to hate me for using the old ETC1.dll, viewer, etc. 

 

If it doesn't close on its own, you must close it yourself ASAP. This means that it's stuck in a file that doesn't have all of the needed chunks and will keep eating disk space with the output file.

[haddockd]

What language is this written in?

 

If its c#.net, you can have the execution stop after a certain amount of time to avoid this issue.

 

Here is a sample code snippet

using System.Timers;


Timer _timer = new Timer();

_timer.Interval = 7000; // Time in milliseconds, so this would be 7 seconds
_timer.Tick += Timer_Tick;
_timer.Start();
convertModels("path1", "path2");

private void convertModels(string fileIn, string fileOut)
{
    converter logic here
}

void Timer_Tick(object sender, EventArgs e)
{
    _timer.Stop();
    MessageBox.Show("There is an issue with the converter. \nPress Enter to stop the process", "Too much space taken", MessageBoxButtons.OK, MessageBoxIcon.Error);
    System.Environment.Exit(0);
}

This would kill the converter if it takes more than 7 seconds. I didnt test it, but it *should* work.

 

A better way is to check the file before processing to ensure it has all the needed chunks.

[Twili]

What language is this written in?

 

If its c#.net, you can have the execution stop after a certain amount of time to avoid this issue.

 

Here is a sample code snippet

using System.Timers;


Timer _timer = new Timer();

_timer.Interval = 7000; // Time in milliseconds, so this would be 7 seconds
_timer.Tick += Timer_Tick;
_timer.Start();
convertModels("path1", "path2");

private void convertModels(string fileIn, string fileOut)
{
    converter logic here
}

void Timer_Tick(object sender, EventArgs e)
{
    _timer.Stop();
    MessageBox.Show("There is an issue with the converter. \nPress Enter to stop the process", "Too much space taken", MessageBoxButtons.OK, MessageBoxIcon.Error);
    System.Environment.Exit(0);
}

This would kill the converter if it takes more than 7 seconds. I didnt test it, but it *should* work.

 

A better way is to check the file before processing to ensure it has all the needed chunks.

'tis a false alarm. The issue happens if you try to batch-run it too many times in a row. Works fine in small doses. Also, C. The code is very ugly!

 

http://pointblank.bz/pb/Nontondo_64/mm2oot.c

[haddockd]

Is there a specific number of times it must run before the issue happens?

 

It may be prudent to simply spawn a thread for each run, one at a time until completion of the job (foreach loop passing a backGroundWorker for each run). That should prevent such issues. Although without the code there is no way to know for sure.

 

Edit: I just realized I may be speaking out of turn because you never asked for my help and here I am throwing it out there. Sorry

[xdaniel]

xdaniel is going to hate me for using the old ETC1.dll, viewer, etc. 

Not "hate", no. Just "strongly dislike"?

 

Nah, just joking. Seriously tho, I guess this is useful for anyone not able to run newer builds of the viewer, for whatever reason. Anyone else:

 

 

Also figured out and fixed some more material-related stuff, like texture wrap modes (mirroring, etc. now works correctly) and source/destination blending factors. Like a lot of the format, this is shared with OoT3D.

[Zeth Ryder]

Would it be possible to get an executable of the build for those who aren't able to compile it? I'm not much of a programmer and only have compiled things on the very rare occasion and its rare they turned out working lol.

[xdaniel]

Would it be possible to get an executable of the build for those who aren't able to compile it? I'm not much of a programmer and only have compiled things on the very rare occasion and its rare they turned out working lol.

I'll post an "unstable" one over the next couple of days. There's still some relatively serious problems, ex. it feeding invalid data to OpenGL on occasion, which I'd still like to fix, even for an unstable build.

 

I know that PwnzLPs posted a build based on the first commit to the repository over on Zelda64.net, but that's a bit outdated by now and not supported by me. Well, as far as I can try to support things anyway... you know what I mean? Anyway, as I said, I'll post one soon-ish.

 

Edit, actually: https://github.com/xdanieldzd/N3DSCmbViewer/releases/tag/v025-unstable - Unstable!

 

Edit 2: Make sure "Options" -> "Enable Skeletal Rendering" is checked for meshes in scenes to be positioned correctly. That'll break actor rendering because it's incomplete, tho.

[Zeth Ryder]

Awesome! Thanks for the build Xdan! Its looking really great!

[kyrin64]

any where I can get the texture rips 

[haddockd]

I would help with the code but I don't know C and am in no hurry to learn it haha. I never really cared for non-OO languages.

[PwnzLPs]

any where I can get the texture rips 

There are no texture rips yet. For now, you'll have to do them yourself. I am working on ripping the textures from every model in the game, but that is going to take some time, because I have to export every model, and then grab their textures.

 

/PwnzLPs

[Zeth Ryder]

Annnnnd just for the hell of it! Also yes I know the fountain is missing the waterfalls and vines.

[kyrin64]

There are no texture rips yet. For now, you'll have to do them yourself. I am working on ripping the textures from every model in the game, but that is going to take some time, because I have to export every model, and then grab their textures.

 

/PwnzLPs

The reason I want to do texture rips and model rips is because I think the game looks really good, and being the 3d modeler that I am, I like to look at game textures and models to see how their constructed, but yeah I guess I'll try to rip them my self 

[xdaniel]

New viewer build: https://github.com/xdanieldzd/N3DSCmbViewer/releases/tag/v025-unstable2 - faster rendering, more stable, some Collada export fixes!

[Twili]

I converted the LZSS decompressor that's part of N3DSCmbViewer to a standalone C program.

#include <stdio.h>

int main(int argc, char **argv)
{
    FILE *arc, *out;
    unsigned int tag,length,decompressedSize,compressedSize;
    unsigned short writeidx=0xFEE,readidx,y;
    unsigned char data[4096],flags,byte,x;
    arc=fopen(argv[1],"rb");
    out=fopen(argv[2],"wb");
    tag=(unsigned int)((fgetc(arc)<<24)|(fgetc(arc)<<16)|(fgetc(arc)<<8)|fgetc(arc));
    if(tag!=0x4C7A5301){printf("This isn't compressed!\n");return -1;}
    fseek(arc,4,SEEK_CUR);
    decompressedSize=(unsigned int)(fgetc(arc)|(fgetc(arc)<<8)|(fgetc(arc)<<16)|(fgetc(arc)<<24));
    compressedSize=(unsigned int)(fgetc(arc)|(fgetc(arc)<<8)|(fgetc(arc)<<16)|(fgetc(arc)<<24));
    fseek(arc,0,SEEK_END);
    length=ftell(arc);
    fseek(arc,0x10,SEEK_SET);
    if(length!=compressedSize+0x10){printf("Size mismatch.\n");return -1;}
    for(y=0;y<4096;y+=1){data[y]=0;}
    while(ftell(arc)<length)
    {
        flags=fgetc(arc);
        for(x=0;x<8;x+=1)
        {
            if((flags&1)!=0)
            {
                byte=fgetc(arc);
                fputc(byte,out);
                data[writeidx]=byte;
                writeidx+=1;writeidx%=4096;
            }
            else
            {
                readidx=fgetc(arc);
                byte=fgetc(arc);
                readidx|=(unsigned short)((byte&0xF0)<<4);
                for(y=0;y<(byte&0xF)+3;y+=1)
                {
                    fputc(data[readidx],out);
                    data[writeidx]=data[readidx];
                    readidx+=1;readidx%=4096;
                    writeidx+=1;writeidx%=4096;
                }
            }
            flags>>=1;
            if(ftell(arc)>=length){break;}
        }
    }
    return 0;
}

[xdaniel]

Here's the original C++ LZSS code by ShimmerFairy (lue) that I based my C# version on: https://github.com/lue/MM3D/blob/master/src/lzs.cpp

 

That aside, not much to report aside from some quite preliminary ZSI support:

 

 

[Twili]

Here's the original C++ LZSS code by ShimmerFairy (lue) that I based my C# version on: https://github.com/lue/MM3D/blob/master/src/lzs.cpp

I knew of that. I couldn't compile it, which is why I rewrote it in C. Thanks anyway. It used at least one Linux gcc header, which annoyed me since it wasn't all set to compile without more work on my end. I looked at your C# version because it was easier to understand.