Transitional Compression

If the ulCompression field is set to BI_TRANS, the data is transitionally compressed using a table of byte transitions or sequences. Values in the data indicate a table position to start at, and the table provides continuing references to other table positions. Transitional compression applies only to eight-bit data, either from an eight-bit palettized image or from a multi-plane image in which each color component is represented in eight bits.

The table consists of up to 256 16-byte rows at the beginning of the data section of the object. Nibbles (half-bytes) in the data section indicate an offset into a table row, at which location is stored the actual byte value. The actual value then becomes the row applicable to the next data nibble. The transitional encoding scheme is described more fully in a separate IBM document.

In transitional compression, the data section is a two-part compound object having the following items:

• A transition table

• The compressed image data

The transition table consists of an integer indicating the table size in bytes and a table of 16-byte rows. The first byte in each row is a row number and the next 15 are transition values. Rows are in descending sequence. The image is compressed according to the following rules:

• Data is in nibbles (half-bytes) or in nibble-pairs (successive half-bytes which may cross a byte boundary).

• The first byte is a nibble-pair. It is the first byte of the image and also the first row number.

• Following a nibble-pair is a series of transition nibbles (1[-15) ended by a terminator (0). Each transition nibble indicates an offset in the current row at which the next byte in the image is found; this value is also the next row number.

• The terminator indicates that the next image byte is not in the table, but instead in the following nibble-pair. This value is also the next row number.

• If the picture has an odd number of nibbles (i.e., it ends in the first half of the last byte), an extra zero nibble is included.