ASE Files (.ASE) Format description
Copyright (C) 2001-2012 by David Capello
----------------------------------------------------------------------

1. References
2. Introduction
3. Header
4. Frames
5. New chunk types
6. File Format Changes

========================================
1. References
========================================

ASE files use Intel (little-endian) byte order.

BYTE            An 8-bit unsigned integer value
WORD            A 16-bit unsigned integer value
DWORD           A 32-bit unsigned integer value
LONG            A 32-bit signed integer value
BYTE[n]         "n" bytes.
RECT            Four LONGs (in the order: x-pos, y-pos, width, heigth)
STRING          length=WORD (how many characters to read next)
                string=BYTE[length]
                The \0 character isn't included.
PIXEL           Pixel format:
                - In RGB images, each pixel have 4 bytes in
                  this order Red, Green, Blue, Alpha.
                - In Grayscale images, each pixel have 2 bytes
                  in the order Value, Alpha.
                - In Indexed images, each pixel uses 1 byte (the index).


========================================
2. Introduction
========================================

The format is much like FLI/FLC files, but with different magic number
and differents chunks.  Also, the color depth can be 8, 16 or 32 for
Indexed, Grayscale and RGB respectively, and the images are more easy
to read (there aren't differences with old frames).  The palette isn't
included for color depths more than 8, but when the sprite is 8 bpp,
the palette is in a FLI color chunk (it could be type=11 or type=4).
See fli.txt for details.

To read the sprite, just do this:
* Read the ASE header (section 3)
* For each frame do (how many frames? the ASE header has that
  information):
  + Read the frame header (section 4)
  + For each chunk in this frame (how many chunks? the frame header
    has that information)
    - Read the chunk (it should be layer information, a cel or a
      palette)


========================================
3. Header
========================================

A 128-byte header (same as FLC/FLI header, but with other magic
number):

DWORD           File size
WORD            Magic number (0xA5E0)
WORD            Frames
WORD            Width in pixels
WORD            Height in pixels
WORD            Color depth (bits per pixel)
                32 bpp = RGBA
                16 bpp = Grayscale
                 8 bpp = Indexed
DWORD           Flags (must be 0)
WORD            Speed (milliseconds between frame, like in FLC files)
                DEPRECATED: You should use the frame duration
                            field from each frame header
DWORD           Set be 0
DWORD           Set be 0
BYTE            Palette entry (index) which represent transparent
                color in all non-background layers (only for Indexed
                sprites).
BYTE[3]         Ignore these bytes
WORD            Number of colors (0 means 256 for old sprites)
BYTE[94]        For future (set to zero)


========================================
4. Frames
========================================

After the header come the "frames" data. Each frame has this little
header of 16 bytes:

  DWORD         Bytes in this frame
  WORD          Magic number (always 0xF1FA)
  WORD          Number of "chunks" in this frame
  WORD          Frame duration (in milliseconds)
  BYTE[6]       For future (set to zero)

Then each chunk format is:

  DWORD         Chunk size
  WORD          Chunk type
  BYTE[]        Chunk data


========================================
5. New chunk types
========================================

Layer Chunk (0x2004)
----------------------------------------

  In the first frame should be a set of layer chunks to determine the
  entire layers layout:

  WORD          Flags:
                  LAYER_IS_READABLE     = 0x0001
                  LAYER_IS_WRITABLE     = 0x0002
                  LAYER_IS_LOCKMOVE     = 0x0004
                  LAYER_IS_BACKGROUND   = 0x0008
  WORD          Layer type (0=normal (image) layer, 1=layer set)
  WORD          Layer child level (see NOTE.1)
  WORD          Default layer width in pixels (ignored)
  WORD          Default layer height in pixels (ignored)
  WORD          DEPRECATED This field is not used anymore.
                Blend mode (always 0 for layer set)
                  BLEND_MODE_NORMAL     = 0
                  BLEND_MODE_DISSOLVE   = 1
                  BLEND_MODE_MULTIPLY   = 2
                  BLEND_MODE_SCREEN     = 3
                  BLEND_MODE_OVERLAY    = 4
                  BLEND_MODE_HARD_LIGHT = 5
                  BLEND_MODE_DODGE      = 6
                  BLEND_MODE_BURN       = 7
                  BLEND_MODE_DARKEN     = 8
                  BLEND_MODE_LIGHTEN    = 9
                  BLEND_MODE_ADDITION   = 10
                  BLEND_MODE_SUBTRACT   = 11
                  BLEND_MODE_DIFFERENCE = 12
                  BLEND_MODE_HUE        = 13
                  BLEND_MODE_SATURATION = 14
                  BLEND_MODE_COLOR      = 15
                  BLEND_MODE_LUMINOSITY = 16
                  BLEND_MODE_COPY       = 17
  BYTE[4]       For future (set to zero)
  STRING        Layer name


Cel Chunk (0x2005)
----------------------------------------

  This chunk determine where to put a cel in the specified
  layer/frame.

  WORD          Layer index (see NOTE.2)
  WORD          X position
  WORD          Y position
  BYTE          Opacity level
  WORD          Cel type
  BYTE[7]       For future (set to zero)

  + For cel type = 0 (Raw Cel):

    WORD        Width in pixels
    WORD        Height in pixels
    PIXEL[]     Raw pixel data: row by row from top to bottom,
                for each scanline read pixels from left to right.

  + For cel type = 1 (Linked Cel):

    WORD        Frame position to link with

  + For cel type = 2 (compressed image):

    WORD        Width in pixels
    WORD        Height in pixels
    BYTE[]      Compressed "Raw Cel" data. Details about the
                compression method:
                http://www.ietf.org/rfc/rfc1951.txt


Mask Chunk (0x2016) DEPRECATED
----------------------------------------

  WORD          X position
  WORD          Y position
  WORD          Width
  WORD          Height
  BYTE[8]       For future (set to zero)
  STRING        Mask name
  BYTE[]        Bit map data (size = height*((width+7)/8))
                Each byte contains 8 pixels (the leftmost pixels are
                packed into the high order bits)


Path Chunk (0x2017)
----------------------------------------

  Never used.


Notes
----------------------------------------

NOTE.1: The child level is used to show the relationship of this
        layer with the last one read, for example:

  Layer name and hierarchy      Child Level
  -----------------------------------------------
  - Background                  0
    `- Layer1                   1
  - Foreground                  0
    |- My set1                  1
    |  `- Layer2                2
    `- Layer3                   1


NOTE.2: The layer index is a number to identify any layer in the
        sprite, for example:

  Layer name and hierarchy      Layer index
  -----------------------------------------------
  - Background                  0
    `- Layer1                   1
  - Foreground                  2
    |- My set1                  3
    |  `- Layer2                4
    `- Layer3                   5


========================================
File Format Changes
========================================

  1) The first change from the first release of the new .ase format,
     is the new frame duration field. This is because now each frame
     can have different duration.

     How to read both formats (old and new one)?  You should set all
     frames durations to the "speed" field read from the main ASE
     header.  Then, if you found a frame with the frame-duration
     field > 0, you should update the duration of the frame with
     that value.