src.mkds_extensions package

Submodules

src.mkds_extensions.kcl_torch module

class src.mkds_extensions.kcl_torch.PrismsTensor(_height: Sequence[float], _pos_i: Sequence[int], _fnrm_i: Sequence[int], _enrm1_i: Sequence[int], _enrm2_i: Sequence[int], _enrm3_i: Sequence[int], _attributes: Sequence[Sequence[int]], device=None)

Bases: PrismsBase

Represents the triangular prisms section of the KCL file.

Each prism is a 0x10 byte structure with the following layout:

Prism Entry Layout (Size: 0x10)

Offset	Type	Name	Description
0x00	f32	height	Prism height from vertex 1 to opposite side
0x04	u16	pos_i	Index of first vertex in positions array
0x06	u16	fnrm_i	Face normal index
0x08	u16	enrm1_i	Edge normal A index
0x0A	u16	enrm2_i	Edge normal B index
0x0C	u16	enrm3_i	Edge normal C index
0x0E	u16	attributes	Collision attributes

_height

Prism heights

Type:

list[float]

_pos_i

Vertex indices

Type:

list[int]

_fnrm_i

Face normal indices

Type:

list[int]

_enrm1_i

Edge normal 1 indices

Type:

list[int]

_enrm2_i

Edge normal 2 indices

Type:

list[int]

_enrm3_i

Edge normal 3 indices

Type:

list[int]

_attributes

Collision attribute flags

Type:

list[int]

property map_2d_shadow

property light_id

property ignore_drivers

property collision_variant

property collision_type

property ignore_items

property is_wall

property is_floor

class src.mkds_extensions.kcl_torch.KCLTensor(data: bytes, prisms: PrismsTensor, positions: Sequence[Sequence[float]], normals: Sequence[Sequence[float]], _positions_offset: int, _normals_offset: int, _prisms_offset: int, _block_data_offset: int, _prism_thickness: float, _area_min_pos: tuple[float, float, float], _area_x_width_mask: int, _area_y_width_mask: int, _area_z_width_mask: int, _block_width_shift: int, _area_x_blocks_shift: int, _area_xy_blocks_shift: int, _sphere_radius: int | None, device=None)

Bases: KCLBase

Represents a KCL (collision) file.

KCL files store simplified model data for collision detection in games such as Mario Kart Wii / DS. They consist of a header, positions, normals, triangular prisms, and octree blocks.

Header Layout

Offset	Type	Name	Description
0x00	u32	positions_offset	Offset to vertex positions section
0x04	u32	normals_offset	Offset to normal vectors section
0x08	u32	prisms_offset	Offset to triangular prisms section
0x0C	u32	block_data_offset	Offset to octree blocks
0x10	f32	prism_thickness	Depth of triangular prism along normal
0x14	Vec3	area_min_pos	Minimum coordinates of model bounding box
0x20	u32	area_x_width_mask	X-axis mask for octree
0x24	u32	area_y_width_mask	Y-axis mask for octree
0x28	u32	area_z_width_mask	Z-axis mask for octree
0x2C	u32	block_width_shift	Octree leaf block size shift
0x30	u32	area_x_blocks_shift	Root child index shift (Y axis)
0x34	u32	area_xy_blocks_shift	Root child index shift (Z axis)
0x38	f32?	sphere_radius	Optional: max sphere radius for collisions

_positions_offset

File offset to position vectors

Type:

int

_normals_offset

File offset to normal vectors

Type:

int

_prisms_offset

File offset to prism data

Type:

int

_block_data_offset

File offset to octree blocks

Type:

int

_prism_thickness

Depth of each prism

Type:

float

_area_min_pos

Minimum coordinates of the collision area

Type:

list[float]

_area_x_width_mask

X-axis mask for octree

Type:

int

_area_y_width_mask

Y-axis mask for octree

Type:

int

_area_z_width_mask

Z-axis mask for octree

Type:

int

_block_width_shift

Octree leaf size shift

Type:

int

_area_x_blocks_shift

Root block child index shift (Y)

Type:

int

_area_xy_blocks_shift

Root block child index shift (Z)

Type:

int

_sphere_radius

Optional maximum sphere radius for collisions

Type:

float or None

_prisms

Parsed prism objects

Type:

Prisms

_positions

List of vertex positions

Type:

list

_normals

List of normal vectors

Type:

list

prism_cls

alias of PrismsTensor

search_triangles(point: tuple[float, float, float] | torch.Tensor, filter_attribute_id: int | None = None)

nearest_triangles(point, n=1, device=None)

classmethod from_file(path: str, device=None)

src.mkds_extensions.nkm_torch module

class src.mkds_extensions.nkm_torch.CPOITensor(data, device=None)

Bases: CPOI

class src.mkds_extensions.nkm_torch.NKMTensor(data, device=None)

Bases: NKM

Module contents