BIFF files

Description

This is a simple archive format, used both to simplify organization of the files, as well as to group together entire areas. Some BIF files are grouped together because of common functionality; for instance, all the .dlg files are gathered together into a single BIF file in Planescape: Torment. Other bif files, such as AR4800.BIF represent the entirety of an area's terrain (i.e. not its containers, actors, or items). These BIF files contain, typically:

at least one .wed file, detailing how the tiles are to be drawn to the screen (i.e. which are animated, which are linked to the opening of which door, etc.) and where walls are (for clipping of actor graphics)
at least one .tis file, which contains all the tile data for the area
a .mos file which is the overhead map of the region
3 or 4 small bitmap files which contain one pixel for each tile needed to cover the region.

The three (or four) bitmaps, are named xxxxxxHT.BMP, xxxxxxLM.BMP, xxxxxxSR.BMP, and (when there are four bitmaps) xxxxxxLN.BMP.

xxxxxxHT.BMP: Height-map, which gives the altitude of each tile cell in the associated wed file
xxxxxxLM.BMP: Light-map, giving the level of illumination (or shadow) of each tile cell on the map. (Called a LUM-map by the developers.)
xxxxxxLN.BMP: When present, this is the same as xxxxxxLM.BMP, but for night-time only. When present, there will also be a second wed file to be used only at night.
xxxxxxSR.BMP: This file has been confirmed to be a "search map", used in the path-finding algorithms

BIFF file versions

V1: Baldur's Gate, Planescape: Torment, Icewind Dale, Baldur's Gate 2
V1 (compressed): Icewind Dale only
BIFC V1.0 (compressed): Baldur's Gate 2 only

BIFF V1

Overall structure

Header
File entries
Tileset entries
Data for the contained files

BIFF V1 Header

Offset	Size (datatype)	Description
0x0000	4 (char array)	Signature ('BIFF')
0x0004	4 (char array)	Version ('V1 ')
0x0008	4 (dword)	Number of file entries
0x000c	4 (dword)	Number of tileset entries
0x0010	4 (dword)	Offset of file entries (from start of file). Tileset entries, if any, immediately follow the file entries.

BIFF V1 File Entries

Offset	Size (datatype)	Description
0x0000	4 (dword)	Resource locator (Note: On disk, only bits 0-13 are matched upon. They are matched against the file index in the "resource locator" field from the KEY file resource entries which claim to exist in this BIFF.
0x0004	4 (dword)	Offset to resource data from start of file
0x0008	4 (dword)	Size (in bytes) of this resource (file).
0x000c	2 (word)	Type of this resource. (See the table of resTypes for a list of which values are applicable to this field
0x000e	2 (word)	Unknown. Part of the previous field? (Or only used in the memory image?)

BIFF V1 Tileset Entries

Offset	Size (datatype)	Description
0x0000	4 (dword)	Resource locator (Note: On disk, only bits 14-19 are matched upon. They are matched against the tileset index in the "resource locator" field from the KEY file resource entries which claim to exist in this BIFF.
0x0004	4 (dword)	Offset to resource data from start of file
0x0008	4 (dword)	Number of tiles in this resource (file).
0x000c	4 (dword)	Size (in bytes) of each tile in this resource (file).
0x0010	2 (word)	Type of this resource. (These resources are always type `0x3eb` -- .tis files, but see the table of resTypes for a list of which values are applicable to this field
0x0012	2 (word)	Unknown. Part of the previous field? (Or only used in the memory image?)

BIFF V1 (compressed)

Overall structure

Header
Compressed file data

This format is idential to the SAV file format, except for the first 8 bytes, which are 'BIF V1.0' for these files, and 'SAV V1.0' for SAV files. The data, once uncompressed, is formatted as a normal BIF file, as seen above. Note that the SAV files can hold more than one compressed file. It is unknown whether this occurs also in CBF files. The compressed format is as follows:

Compressed BIFF V1

Offset	Size (datatype)	Description
0x0000	4 (char array)	Signature ('BIF ')
0x0004	4 (char array)	Version ('V1.0')
0x0008	4 (dword)	Length of filename
0x000c	variable (ASCIIZ char array)	Filename (length specified by previous field)
sizeof(filename)+0x0010	4 (dword)	uncompressed data length
sizeof(filename)+0x0014	4 (dword)	compressed data length
sizeof(filename)+0x0018	variable (raw data)	compressed data (length in bytes specified by previous field)

BIFC V1.0 (compressed)

Overall structure

Header
zlib-compressed blocks

The file is broken up into blocks (typically 8192 bytes apiece), which are compressed separately. The blocks are stored one right after another, immediately after the header. Each block consists of an uncompressed size and a compressed size, and then the data, compressed by zlib. Here is some C code which uses zlib to decompress BIFC files.

BIFC V1.0 header

Offset	Size (datatype)	Description
0x0000	4 (char array)	Signature ('BIFC')
0x0004	4 (char array)	Version ('V1.0')
0x0008	4 (dword)	Uncompressed BIF size

BIFC V1.0 compressed blocks

Offset	Size (datatype)	Description
0x0000	4 (dword)	Decompressed size
0x0004	4 (dword)	Compressed size
0x0008	varies (bytes)	Compressed data

[ back to index ]