diff options
Diffstat (limited to 'lss16.5')
| -rw-r--r-- | lss16.5 | 61 |
1 files changed, 61 insertions, 0 deletions
@@ -0,0 +1,61 @@ +.TH LSS16 5 LIBLSS16 +.SH NAME +lss16 \- SYSLINUX \(dqad hoc\(dq 16-colour, 6-bit channel-depth image file format LSS16 + +.SH DESCRIPTION +LSS16 is run-length encoding compressed, lossless an image file format +using a 16-colour palette and sRGB 6-6-6. LSS16 is designed for bootloader splash +images which may be displayed on low-end legacy devices with limited image quality +capabilities. LSS16 does not support transparency. +.PP +LSS16 file consists how two parts: a 56 byte header followed by the pixel encoding. +.PP +Header format: +.RS +4-byte magic number: 3Dh, F3h, 13h, 14h. + +Image width (number of pixels horizontally): unsigned 16-bit integer, little endian. + +Image height (number of pixels horizontally): unsigned 16-bit integer, little endian, +however the most significant bit most be cleared. + +Colour palette: 16 RGB-tuples, each tuple is 3 bytes: red, green, blue. +The two most significant bits in each byte must be cleared. +.RE +.PP +Pixel encoding format: +.RS +The pixels are encoded, row by row, from the top down, in a sequence of nibble. Each +row is encoded from left to right, and odd-nibble rows are padded with a zero-nibble. +Each encoded byte, is split into two nibbles; the lower part of the byte makes up the +first nibble, and the higher part of the byte makes up the second nibble. + +At the beginning of each row, the previous pixel is assumed to use the 0-indexed colour. + +If a nibble as the value of the the previous pixel colour index, a run-length is encoded, +otherwise the next pixel has the colour with the index encoded in the nibble. + +If the nibble encodes a run-length, it can either be 0 or the number of times the +previous pixel is repeated (excluding the already existing pixels). If the nibble +is 0, the next two nibbles, encodes in little-endian, a value +.IR N , +and the previous pixel is repeated +.IR N +16 +times (excluding the already existing pixels). +.RE +.PP +A 6-bit colour-channel +value +.I N +is mapped to 8-bits with the function +.I (N*255+31)/63 +(linear scaling). +.PP +LSS16 image files usually have the filename suffix +.RB \(dq .lss \(dq +or +.RB \(dq .16 \(dq. + +.SH SEE ALSO +.BR liblss16 (7), +.BR syslinux (1) |