CGFX is a container format used to store graphics resources. It can contain 3D models, textures and animation data.

CGFX

CGFX header :

Offset Length Description
0x0 0x4 Magic "CGFX"
0x4 0x2 Byte order mark: FFFE (little endian) or FEFF (big endian)
0x6 0x4 CGFX header size
0xA 0x2 ?
0xC 0x4 File size (bytes)
0x10 0x4 Number of entries

A typical CGFX file contains two main entries, beginning directly after the CGFX header: DATA and IMAG.

DATA

DATA contains a list of DICT references.

DATA header (for N = 0..15) :

Offset Length Description
0x0 0x4 Magic "DATA"
0x4 0x4 DATA Size (in bytes)
0x8 +(N*8) 0x4 Number of entries in DICT N
0xC +(N*8) 0x4 Offset (self-relative) to DICT N

The DATA header contains the entry counts and offsets for each DICT entry. The number of entries can vary (probably based on the version?), but are always in the following order. Any unused entries are zeroed.

Typical entries:

N Type
0 Models
1 Textures
2 LUTS (Material/Color/Shader look-up tables?)
3 Unknown
4 Unknown
5 Cameras
6 Lights
7 Fog
8 Environments
9 Skeleton animations
10 Texture animations
11 Unknown animations
12 Unknown
13 Unknown
14 Unknown
15 Unknown

DICT

DICTs are generic structures used to store values (and associate them to a key ?). A DICT header is 0x1C bytes long.

DICT header :

Offset Length Description
0x0 0x4 Magic "DICT"
0x4 0x4 DICT size (in bytes)
0x8 0x4 Number of entries
0xC 0x10 ?

DICT entry:

Offset Length Description
0x0 0x4 ?
0x4 0x2 ?
0x6 0x2 ?
0x8 0x4 Offset (self-relative) to symbol
0xC 0x4 Value (often offsets)

CMDL

CMDL is used to describe a 3D model.

CMDL Header :

Offset Length Description
0x0 0x4 Flags (bit 7: hasSkeletonSobj)
0x4 0x4 Magic "CMDL"
0x8 0x4 ?
0xC 0x4 Offset (self-relative) to model name
0x10 0x18 ?
0x28 0x4 Number of entries in Animation Types DICT
0x2C 0x4 Offset (self-relative) to Animation Types DICT
0x30 0xC Global scale vector (3 floats : x, y, z)
0x3C 0x18 ?
0x54 0x30 Matrix 1
0x84 0x30 Matrix 2
0xB4 0x4 Number of Vertex Info SOBJ entries
0xB8 0x4 Offset (self-relative) to Vertex Info SOBJ list
0xBC 0x4 Number of MTOB DICT entries
0xC0 0x4 Offset (self-relative) to MTOB DICT
0xC4 0x4 Number of Vertex Info SOBJ entries
0xC8 0x4 Offset (self-relative) to Vertex Info SOBJ list
0xCC 0x4 Number of Unknown DICT entries
0xD0 0x4 Offset (self-relative) to Unknown DICT
0xD4 0xC ?
0xE0 0x4 Skeleton Info SOBJ offset (self-relative) [only present if flag bit 7 is set]
0xB8+[0xB8] 0x4*N Vertex Info SOBJ self-relative offset list

A CMDL section refers to outside data; it can not be considered separately from the rest of the CGFX file. The second DICT in the CMDL section contains offsets to MTOB objects.

SOBJ

SOBJ structures can be used to describe 3D objects that are part of the model. If such is the case then they will follow this structure :

Offset Length Description
0x0 0x4 Flags (bit 4: model; bit 1: skeleton)
0x4 0x4 Magic "SOBJ"
0x8 0x4 ?
0xC 0x4 Unknown symbol offset (self-relative)
0x10 0xC ?
0x1C 0x4 Offset (self-relative) to Unknown1 (appears to hold array of floats) ?
0x20 0xC Mesh position offset (X/Y/Z floats)
0x2C 0x4 Face groups count
0x30 0x4 Offset (self-relative) to face groups offset array
0x34 0x4 ?
0x38 0x4 Vertex groups count
0x3C 0x4 Offset (self-relative) to vertex groups offset array
0x40 0x4 Unknown offset (self-relative) ?

Face groups:

Offset Length Description
0x0 0x4 Bone groups count
0x4 0x4 Offset (self-relative) to UInt32 bone group IDs array
0x8 0x4 ?
0xC 0x4 Unknown2 count
0x10 0x4 Offset (self-relative) to Unknown2 offset array

Unknown2:

Offset Length Description
0x0 0x4 Face group descriptor count
0x4 0x4 Offset (self-relative) to face array descriptors offset array
0x8 0x4 Unknown3 count
0xC 0x4 Offset (self-relative) to UInt32 Unknown3 array
0x10 0x8 ?

Face array descriptor:

Offset Length Description
0x0 0x4 Flags (bit 1: vertex index format: 0=byte, 1=short)
0x4 0x4 ?
0x8 0x4 Vertex index array size (in bytes)
0xC 0x4 Offset (self-relative) to vertex index array

Vertex groups come in a number of different formats. Typically the first vertex group entry is of format 0x40000002 and contains the actual vertex array.

Vertex group format 0x40000002:

Offset Length Description
0x0 0x4 Flags (0x40000002)
0x4 0x4 ?
0x8 0x4 ?
0xC 0x4 ?
0x10 0x4 ?
0x14 0x4 Vertex array size (in bytes)
0x18 0x4 Offset (self-relative) to vertex array
0x1C 0x4 ?
0x20 0x4 ?
0x24 0x4 Vertex stride/size in bytes (see below)
0x28 0x4 Unknown3 count
0x2C 0x4 Offset (self-relative) to component declaration offset array

Each mesh's primary vertex group contains an array of vertex component declaration objects, defining the parameters for each of a vertex's components.

Vertex component declaration:

Offset Length Description
0x0 0x4 Flags (0x40000001)
0x4 0x4 Vertex component type (see below)
0x8 0x4 ?
0xC 0x4 ?
0x10 0x4 ?
0x14 0x4 ?
0x18 0x4 ?
0x1C 0x4 ?
0x20 0x4 ?
0x24 0x1 Component data type (see below)
0x25 0x1 ?
0x26 0x1 ?
0x27 0x1 ?
0x28 0x4 Number of values in this component (e.g. XYZ->3, UV->2)
0x2C 0x4 Multiplier for this component's values (float)
0x30 0x4 Position of this component within vertex stride

Observed sample vertex format sizes:

Vertex stride Description
0xA X (short), Y (short), Z (short), U (short), V (short)
0xC X (float), Y (float), Z (float)
0x10 X (short), Y (short), Z (short), ... ?
0x14 X (float), Y (float), Z (float), U (float), V (float)
0x18 X (float), Y (float), Z (float), Unk (u32?), U (float), V (float)
0x20 X (float), Y (float), Z (float), NX (float), NY (float), NZ (float), U (float), V (float)
0x28v1 X (float), Y (float), Z (float), NX (float), NY (float), NZ (float), U (float), V (float), (local) Bone IDs (4*u8), Bone weights (4*u8)
0x28v2 X (float), Y (float), Z (float), NX (float), NY (float), NZ (float), U (float), V (float), Unk1 (u32), (local) Bone IDs (2*u8), Bone weights (2*u8)

Vertex formats with bone data support multiple bone assignment. In this case, the sum of all bone weights is 0x64.

TXOB

TXOBs are contained within MTOBs. They can describe textures; if such is the case, then their structure is as follows :

Offset Length Description
0x0 0x4 Flags
0x4 0x4 Magic "TXOB"
0xC 0x4 Offset (self-relative) to symbol
0x18 0x4 Texture height
0x1C 0x4 Texture width
0x28 0x4 Mipmap levels
0x34 0x4 Texture format ID (see table below)
0x3C 0x4 Texture height (?)
0x40 0x4 Texture width (?)
0x44 0x4 Texture data size
0x48 0x4 Texture data offset (self-relative)
Texture format ID Description
0x0 RGBA8
0x1 RGB8
0x2 RGBA5551
0x3 RGB565
0x4 RGBA4
0x5 LA8
0x6 HILO8
0x7 L8
0x8 A8
0x9 LA4
0xA L4
0xB A4 ?
0xC ETC1 (see notes below)
0xD ETC1A4 ?

Every texture format has its texture data divided into 8x8 tiles. See SMDH for more information. ETC1 is a compressed texture format which compresses blocks of 4x4 pixels into u64s. These u64 are traditionally stored in big endian; however, nintendo's implementation stores them in little endian. ETC1 textures are stored in 8x8 tiles; decompressed 4x4 therefore have to be organized accordingly. See [1] for implementation example.

LUTS

Appears to contain color lookup tables possibly for use with shaders.

LUTS Header:

Offset Length Description
0x0 0x4 Magic "LUTS"
0x4 0x2 Seems to adhere to powers of 2 (width/height/flags?)
0x6 0x2 Seems to adhere to powers of 2 (width/height/flags?)
0x8 0x4 ?
0xC 0x8 all zeroes ?
0x14 0x4 ?
0x18 0x4 Offset to DICT (self-relative) ?

All observed instances have an otherwise unreferenced DICT section immediately afterward (the last LUTS value being a 0x4, which may describe the relative position of that DICT), which appears to describe material specularity.

Skeleton data

Skeleton data is stored in an array. Each entry is 0xE0 bytes in length and organized this way :

Offset Length Description
0x0 0x4 Offset (self relative) to name symbol
0x4 0x4 ?
0x8 0x4 Joint ID
0xC 0x4 Parent joint ID
0x10 0x4 Signed offset (self-relative) to parent joint
0x2C 0xC Angle vector (floats, x, y, z)
0x38 0xC Position vector (floats, x, y, z)
0x44 0x30 Transformation matrix (4x3)
0x74 0x30 Identity matrix ? (4x3)

Each entry stores the joint transformation data twice; once as angle/position vectors and once as a transformation matrix. Each entry also stores a second matrix which appears to always be identity. (?)

CANM

CANMs are used to store skeletal animation data.

Links