AMD_compressed_ATC_texture

GL_AMD_compressed_ATC_texture

Aaftab Munshi
Benj Lipchak

Maurice Ribble, Qualcomm (mribble 'at' qualcomm.com)

Please contact AMD regarding any intellectual property questions/issues 
associated with this extension.

Complete.

Last Modified Date: February 26, 2008
Revision: 7

OpenGL ES Extension #40

Written based on the wording of the OpenGL ES 1.1 specification.

This extension enables support for ATC compressed texture formats.  ATC is 
AMD's proprietary compression algorithm for compressing textures for 
handheld devices to save on power consumption, memory footprint and 
bandwidth.

Three compression formats are introduced:

- A compression format for RGB textures.
- A compression format for RGBA textures using explicit alpha encoding.
- A compression format for RGBA textures using interpolated alpha encoding.

None.

Accepted by the <internalformat> parameter of CompressedTexImage2D and
CompressedTexImage3DOES.

    ATC_RGB_AMD                        0x8C92
    ATC_RGBA_EXPLICIT_ALPHA_AMD        0x8C93
    ATC_RGBA_INTERPOLATED_ALPHA_AMD    0x87EE

None.

Add to Table 3.17:  Specific Compressed Internal Formats

    Compressed Internal Format         Base Internal Format
    ===============================    ====================
    ATC_RGB_AMD                        RGB
    ATC_RGBA_EXPLICIT_ALPHA_AMD        RGBA
    ATC_RGBA_INTERPOLATED_ALPHA_AMD    RGBA


Add to Section 3.8.3, Alternate Image Specification

The details of these formats is not disclosed, so refer to AMD's 
Compressonator tool in order to encode your textures offline:
http://ati.amd.com/developer/compressonator.html

OpenGL ES requires that all mip-levels have to be specified when
CompressedTexImage2D is called to load a palettized texture.  In this case 
the "level" value is either 0 or a negative number to indicate the number of
mip-levels.  This rule is not followed when loading an ATC texture.  The 
"level" number in this case represents the actual mip-level number,
so a negative number will raise an error.

Compressed texture images stored using the ATC compressed image formats are
represented as a collection of 4x4 texel blocks.

ATC_RGB_AMD Format
==================

This format compresses blocks of source texels down to 4 bits per texel.
Assuming 8-bit component source texels, this represents a 8:1 compression
ratio.  This is the best format to use when no alpha channel is needed.

ATC_RGBA_EXPLICIT_ALPHA_AMD Format
==================================

This format compresses blocks of source texels down to 8 bits per texel.
Assuming 8-bit component source texels, this represents a 4:1 compression
ratio.  This is generally the best format to use when alpha transitions are
sharp.

ATC_RGBA_INTERPOLATED_ALPHA_AMD Format
======================================

This format compresses blocks of source texels down to 8 bits per texel.
Assuming 8-bit component source texels, this represents a 4:1 compression
ratio.  This is generally the best format to use when alpha transitions are
gradient.

The size in bytes of a compressed ATC_RGB_AMD image is:

((width_in_texels+3)/4) * ((height_in_texels+3)/4) * 8

The size in bytes of a compressed ATC_RGBA_EXPLICIT_ALPHA_AMD or
ATC_RGBA_INTERPOLATED_ALPHA_AMD image is:

((width_in_texels+3)/4) * ((height_in_texels+3)/4) * 16

INVALID_OPERATION is generated by TexImage2D, TexSubImage2D, 
CompressedTexSubImage2D, or CopyTexSubImage2D if <internalformat> or 
<format> is ATC_RGB_AMD, ATC_RGBA_EXPLICIT_ALPHA_AMD, or 
ATC_RGBA_INTERPOLATED_ALPHA_AMD.

The queries for NUM_COMPRESSED_TEXTURE_FORMATS and 
COMPRESSED_TEXTURE_FORMATS include ATC_RGB_AMD, ATC_RGBA_EXPLICIT_ALPHA_AMD,
and ATC_RGBA_INTERPOLATED_ALPHA_AMD.

09/06/2012    Maurice Ribble   Update contact and add block size data.
02/26/2008    Benj Lipchak     Throw INVALID_OPERATION on subimage updates.
10/24/2007    Benj Lipchak     Add interpolated alpha RGBA format.
09/24/2007    Jon Leech        Assign extension number.
09/05/2007    Benj Lipchak     Cosmetic changes.
08/01/2007    Benj Lipchak     Publication readiness.
07/07/2005    Aaftab Munshi    Fixed typos.
09/15/2004    Aaftab Munshi    Created the document.