Rainbow Six: Siege Models Thread

[MaZTeR]

Update:
https://www.dropbox.com/sh/b2cuse4hp90y ... qTfja?dl=0
Here's the first variant of an extraction tool for Rainbow Six: Siege.

Has anyone else been able to rip/extract models from that game? The only models I've seen are only in https://facepunch.com/showthread.php?t= ... st51027928 and since that, no other models have been released.

I can provide some files from the game if anyone wants to reverse-engineer them.

[zeyd6796]

~~Need a update for this topic~~

We have used ninja ripper x64 dx11/9 on the cracked game to extract data to import, but mesh files (.rip) cannot be open properly.

ninja_importer v1.3, rip2obj, a blender .rip script and Noesis have been tried, but the structure of the meshes are just inappropriate.

Here is the source: https://facepunch.com/showthread.php?t= ... p=51269755

There are some .rip and dds samples which are ripped from the game on the topic:
https://yadi.sk/d/2EtUiNFfxhqL3

What can be done, any idea?

[shakotay2]

There are some .rip and dds samples which are ripped from the game on the topic:
https://yadi.sk/d/2EtUiNFfxhqL3

What can be done, any idea?

learn the 3D basics
A quick glance into those *.rip files reveals that there's (apart from some .dds texture names)
only face indices contained - no vertices (zero bytes, 00 00 00 00..., though).

So how should that work?

[zeyd6796]

Thanks for the quick reply.
The samples were parepared to diagnose or to understand what happend when they were ripped. Then we can change the paremeters or methods to obtain correct data. So, the data structure or algorithm of the game is just inappropriate for ninja ripper. And there is no current archive extractor. Just Intel GPA to rip... Right?

[shakotay2]

I don't really care for "war games", so dunno, but I found a rip file with vertices
so Ninja Ripper doesn't seem to be totally "off track":

Mesh_0094-rip.JPG

[MaZTeR]

~~Need a update for this topic~~

We have used ninja ripper x64 dx11/9 on the cracked game to extract data to import, but mesh files (.rip) cannot be open properly.

ninja_importer v1.3, rip2obj, a blender .rip script and Noesis have been tried, but the structure of the meshes are just inappropriate.

Here is the source: https://facepunch.com/showthread.php?t= ... p=51269755

There are some .rip and dds samples which are ripped from the game on the topic:
https://yadi.sk/d/2EtUiNFfxhqL3

What can be done, any idea?

Didn't realize this thread got bumped.

Well, I tried Ninjaripper and it doesn't work on the legit version. However, I'm not sure are there cracked versions of the game from the latest version with the Skull Rain items.

As for other programs, I'm not sure do they work. I tried Archive_Next and it just crashes every time I try to open a file from the R6 game folder.

[MaZTeR]

So is this dead?

[Gh0stBlade]

The problem is a Forge unpacker is required, if someone makes one, I'll work on the 3D models.

Cheers.

[MaZTeR]

The problem is a Forge unpacker is required, if someone makes one, I'll work on the 3D models.

Cheers.

There is Archive_Next, but the developer of that tool has gone dormant so there's no R6 support for it.... :/

[Tushkan]

I have python script that unpacks current versions of forge. They switched to zstandard to pack data, but other than that structure is not much of a difference. Another bugger is that they seem to hash all strings and store hashes instead of actual names to prevent data mining. I was hoping to inspect the game process with debugger to see how they threat those hashes, but legal exe is virtualized and has attach protection (x64dbg throws an error upon attach, on launch it hits the debug warning or BE warning). I'm still too dumb to break through such protection measures, but I have unpacked samples. So if anyone is interested, I can provide my data. They also have .depgraphbin files that hold \x02 at the beginning (magic number?) and then uint64_hash1, uint64_hash2 and some uint32 (class?, relation type?), hashes match the ones that can be found in names section, describing each file inside a given forge package. I assume this is one of ways to tie up needed resources together.

[Jojje]

I have python script that unpacks current versions of forge. They switched to zstandard to pack data, but other than that structure is not much of a difference. Another bugger is that they seem to hash all strings and store hashes instead of actual names to prevent data mining. I was hoping to inspect the game process with debugger to see how they threat those hashes, but legal exe is virtualized and has attach protection (x64dbg throws an error upon attach, on launch it hits the debug warning or BE warning). I'm still too dumb to break through such protection measures, but I have unpacked samples. So if anyone is interested, I can provide my data. They also have .depgraphbin files that hold \x02 at the beginning (magic number?) and then uint64_hash1, uint64_hash2 and some uint32 (class?, relation type?), hashes match the ones that can be found in names section, describing each file inside a given forge package. I assume this is one of ways to tie up needed resources together.

Very interesting, so you can get the files, but the names are hashed?

[Tushkan]

Yes, Jojje. I have the extracted data with fancy looking file names. And me and my friend have extracted a couple of game models, like Thermite's elite mesh for example. But there are some peculiar versions of a mesh file which I can't seem to fully reverse. So I'm asking for a community's help. Here is some of the extracted data https://www.dropbox.com/sh/b2cuse4hp90y ... qTfja?dl=0 It includes a bunch of models (some are operators, their heads, weapons etc) in game's native format plus a README of currently known data plus some logs. Where I need help is with the packed vertex data (marked as Qc and Dc in the README). If anyone can help me crack how it's packed, that would be great.

Here are a couple of screenshots just to draw your attention =)

[Tushkan]

Ok, nevermind, I figured it out. It turned out to be 4 int16. Three for x,y,z and last is scaler.

[Jojje]

Checking in! Has any progress been made since last post..?

[Tushkan]

Yep, there has been. I unpacked all types of meshes and seem to have found uv's block. There are some questionable blocks tho. And the biggest issue is there are no asset names. I have also found the files that tie an asset together (mesh or several meshes, textures) but it has no link to the skeleton which is odd. And an absence of proper names makes it impossible to pinpoint the exact asset you would want to find.