Images and Colour Models Technical Guide
Section 3 - IMAGES

3.1 An <Image>'s Place Is...

Image data is represented in the DRM by <Image> instances. As one can see by examining the documentation of the <Image> class, <Image> instances can appear in a transmittal only within the context of some <Image Library> instance.

Question:
Since an <Image> instance cannot be a component of anything except an <Image Library>, how is it used for texture mapping within a transmittal?
Answer:
The case of <Images> is an instance of the general rule of how <Library> components are instanced within transmittals: a specific class exists in the DRM ( <Image Mapping Function>, in this case) which describes how the <Library> component is referenced in a particular instance. <Image Mapping Function> is covered later on in Section 3 of this document, after the basics have been discussed.

 
3.2 Anatomy of an <Image>
3.2.1 The Usual Suspects: Fields

The most frequently asked question that new users ask when they first begin studying the <Image> class (and which is captured as a FAQ within the class' documentation) is (paraphrased):

Question:
I don't see a field in the <Image> class for the texels; where are they?
Answer:
The actual texels of an <Image> instance are hidden by the API implementation being used to provide the <Image>. To create the texels of an <Image>, a data provider shall use the Transmittal Access Level 0 API function SE_PutImageData(). To retrieve the texels of an <Image>, a consumer shall use the Transmittal Access Level 0 API function SE_GetImageData().
These two functions accept and retrieve the texels of an <Image> instance as an array of bytes, using the fields of the <Image> to determine how many texels are present and the structure of each.

The level_count and mip_extents_array fields, together with data_is_3D, indicate how many texels are present in the <Image>. The data_is_3D field indicates whether a 3D or 2D image is being represented, while the mip_extents_array (which has level_count entries) specifies the horizontal, vertical, and z dimensions of each MIP level in the <Image>. Mipmapping is discussed in detail below; for a non-mipmapped image, level_count is 1 and the single entry in mip_extents_array gives the dimensions of the <Image>. Note that for a 2D <Image>, the z field of each mip_extents_array entry is set to 1.

The remaining fields of <Image> can be grouped into those which are needed to specify the structure of an individual texel, and those which are needed to interpret the image data when it is retrieved. The image_signature, bits_of_xxx, and min / max_value_of_xxx fields fall into the first category, while the remaining fields fall into the second category.

3.2.2 Image Signatures
3.2.2.1 Overview

The image_signature of an <Image> instance specifies the structure of an individual texel, including which of the bits_of_xxx, min_value_of_xxx, and max_value_of_xxx fields are applicable. Essentially, the image_signature indicates how many components are in an individual texel, what they mean, and which bits_of_xxx / min_value_of_xxx / max_value_of_xxx fields are therefore applicable; the bits_of_xxx and min / max fields themselves specify the size and ranges of the individual components of a texel. The individual image signatures are discussed in detail in the rest of this section.

3.2.2.2 LUMINANCE

If image_signature is set to SE_IMAGSIG_LUMINANCE, then each texel in the <Image> consists of a luminance value, representing the luminance (also called intensity) of the texel. A luminance of 0 indicates that the texel is black, while a luminance of 1 indicates that the texel is white; all intermediate values are shades of grey. An <Image> with this signature may be called a luminance <Image>, intensity <Image>, or a greyscale <Image>. The effect is that of a "black and white" television.

A LUMINANCE <Image> can be composited with a compatible ALPHA image to produce a SE_IMAGSIG_LUMINANCE_AND_ALPHA <Image>. An example of compositing SE_IMAGSIG_LUMINANCE <Images> is the case of large areas of terrain <Polygons>, where the same textures for dirt are repeated over and over. To avoid creating a "quilt" effect on the terrain, a luminance <Image> can be composited with the terrain (using different offsets at different locations to make the effect appear random) to "dirty" the textures and give the terrain a more realistic appearance.

FURTHER CONSTRAINTS:

  1. bits_of_luminance shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx fields) shall be zero.

3.2.2.3 COLOUR_COORDINATE_1

Used to indicate that each texel in the <Image> consists of 1 colour component, namely the first for its colour model (G for RGB, M for CMY, or S for HSV). The first colour component of the colour model shall be the only value in the texel.

Note that a COLOUR_COORDINATE_1 <Image> can be composited with compatible COLOUR_COORDINATE_2 and COLOUR_COORDINATE_3 <Images> of the same dimensions to produce a 123COLOUR <Image>. (These can also be composited with a compatible alpha map to produce a 123COLOUR_ALPHA map.)

FURTHER CONSTRAINTS:

  1. bits_of_colour_coordinate_1 shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) shall be zero.
  2. When an object has K <Image Mapping Functions> providing texture mapping information, at most one of the K <Image Mapping Functions> can be associated with a COLOUR_COORDINATE_1 <Image>. (Multiple COLOUR_COORDINATE_1 <Images> produce an undefined result.)
  3. EXAMPLES:

    1. An RGB texel shall have red as the texel value.
    2. An CMY texel shall have cyan as the texel value.
    3. An HSV texel shall have hue as the texel value.

3.2.2.4 COLOUR_COORDINATE_2

See SE_IMAGSIG_COLOUR_COORDINATE_2's documentation.

3.2.2.5 COLOUR_COORDINATE_3

See SE_IMAGSIG_COLOUR_COORDINATE_3's documentation.

3.2.2.6 ALPHA

If image_signature is set to SE_IMAGSIG_ALPHA, then each texel in the <Image> consists of an alpha value, representing the coverage of the texel. An alpha of 0 indicates that the texel is transparent, while an alpha of 1 indicates that the texel constitutes part of the important content of the <Image>. An <Image> with this signature may be called an alpha channel or an alpha map. For further details, see [FOLEY_VAN_DAM] Section 17.6, "Image Compositing".

Note that an alpha map can be composited with:

FURTHER CONSTRAINTS:
  1. bits_of_alpha shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) fields shall be zero.
  2. When an object has K <Image Mapping Functions> providing its texture mapping information, at most one of the K <Image Mapping Functions> can legally provide an alpha map. (Multiple alpha maps produce an undefined result.)

 
3.2.2.7 BUMP_MAP_HEIGHT

Used to indicate that the <Image> represents a 2-D height bump map (or just height map), specifying information used to modify the surface normals of a smooth surface. When used with a ray-tracing technique, bump mapping introduces variations in intensity across the surface, so that it simulates a rough, wrinkled, or dimpled surface (e.g., the surface of the ocean).

A UV type of bump map can be derived from a height bump map by computing the texel gradients in both the X and Y directions.

Rather than manipulating the colour of a flat surface, bump mapping modifies the physical texture of the surface. For a description of bump mapping, see [WATT].

FURTHER CONSTRAINTS:

  1. bits_of_bump_map_height shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) fields shall be zero.
  2. When an object has K <Image Mapping Functions> providing its texture mapping information, at most one of the K <Image Mapping Functions> can legally provide a bump map. (Multiple bump maps produce an undefined result.)
3.2.2.8 EDCS_CLASSIFICATION_CODE

Used to indicate that each texel in the <Image> consists of 1 value, an EDCS Classification Code (ECC).

CONSTRAINTS:

  1. The size per texel shall be the size of EDCS_Classification_Code.
  2. Since no bits_of, min_value, or max_value fields are needed for this signature, all bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx fields) shall be zero.

3.2.2.9 LUMINANCE_AND_ALPHA

Used to indicate that the <Image> is (functionally) a composite of a luminance <Image> and an alpha <Image> (see SE_IMAGSIG_ALPHA, SE_IMAGSIG_LUMINANCE). Each texel consists of an intensity value followed by an alpha value. No other ordering is possible with this signature.

A LUMINANCE_AND_ALPHA <Image> can be down-sampled to produce an ALPHA <Image> and a LUMINANCE <Image>.

FURTHER CONSTRAINTS:

  1. bits_of_luminance + bits_of_alpha shall equal the size per texel (in bits); all other bits_of_xxx fields shall be zero.
  2. When an object has K <Image Mapping Function> instances providing its texture mapping information, at most one of of the K <Image Mapping Functions> instances can legally provide a luminance & alpha map. (Multiple alpha maps produce an undefined result.)

3.2.2.10 123COLOUR

Used to indicate that each texel in the <Image> consists of 3 colour components (RGB, CMY, or HSV). The first colour component of the colour model shall be the first value in the texel, the second colour component of the colour model shall be the second value in the texel, and the third colour component of the colour model shall be the third value in the texel. No other ordering is possible with this signature. Note that a 123COLOUR <Image> can be down-sampled to produce a COLOUR_COORDINATE_1, COLOUR_COORDINATE_2, or COLOUR_COORDINATE_3 <Image>.

FURTHER CONSTRAINTS:

  1. bits_of_colour_coordinate_1 + bits_of_colour_coordinate_2 + bits_of_colour_coordinate_3 shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) shall be zero.
  2. When an object has K <Image Mapping Functions> providing texture mapping information, at most one of the K <Image Mapping Functions> can be associated with a 123 colour map <Image>. (Multiple 123 colour maps produce an undefined result.)
EXAMPLES:
  1. An RGB texel shall have red as the first value in the texel, green next and finally blue.
  2. A CMY texel shall have cyan as the first value in the texel, magenta next, and finally yellow.
  3. An HSV texel shall have hue as the first value in the texel, saturation next and finally brightness value.
3.2.2.11 123COLOUR_ALPHA

Used to indicate that each texel in the <Image> consists of 3 colour components (RGB, CMY, or HSV) and an alpha value. The first colour component of the colour model shall be the first value in the texel, the second colour component of the colour model shall be the second value in the texel, the third colour component of the colour model shall be the third value in the texel, and the alpha value shall be the last value in the texel. No other ordering is possible with this signature.

Note that a 123COLOUR_ALPHA <Image> can be down-sampled to produce a COLOUR_COORDINATE_1, COLOUR_COORDINATE_2, COLOUR_COORDINATE_3, and / or ALPHA <Image>.

FURTHER CONSTRAINTS:

  1. bits_of_colour_coordinate_1 + bits_of_colour_coordinate_2 + bits_of_colour_coordiante_3 + bits_of_alpha shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) shall be zero.
  2. When an object has K <Image Mapping Functions> providing texture mapping information, at most one of the K <Image Mapping Functions> can be associated with a 123 colour & alpha map <Image> (Multiple 123 colour and alpha maps produce an undefined result.)

EXAMPLES:

  1. An RGBA texel shall have red as the first value in the texel, green next, then blue, and finally alpha.
  2. A CMYA texel shall have cyan as the first value in the texel, magenta next, then yellow, and finally alpha.
  3. An HSVA texel shall have hue as the first value in the texel, saturation next, then brightness value, and finally alpha.
3.2.2.12 ONE_MATERIAL

Used to indicate that each texel in the <Image> consists of 1 value, an index into the <Property Tables> referenced by this <Image>. These <Property Tables> describe the material. Normally, a <Property Table Reference> is used to find the corresponding <Property Table>, find which <Axis> is referred to, and which hash value measurement along that <Axis> is being referenced. The bits_of_material_1 field is used in place of the <Property Table Reference>' index_on_axis fields.

See the examples for the <Image> class.

CONSTRAINTS:

  1. The <Image> shall have at least one <Property Table Reference>. (The data producer may have as many <Property Table References> as desired, as long as there is at least one, and bits_of_material_# will handle them.)
  2. All the <Property Tables> being referred to by the <Property Table Reference> shall be of the same size, since the material reference corresponds to all of them.
  3. bits_of_material_1 shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) shall be zero.
3.2.2.13 TWO_MATERIALS

Used to indicate that each texel in the <Image> represents a linear combination of 2 materials in the <Property Tables> referenced by this <Image>. That is, each texel consists of 3 values: 2 indexes into the <Property Tables> referenced by this <Image>, and the percentage (an integer, 0 - 100%) of material 2. These <Property Tables> describe the materials. Normally, a <Property Table Reference> is used to find the corresponding <Property Table>, find which <Axis> is referred to, and which hash value measurement along that <Axis> is being referenced. The bits_of_material1 and bits_of_material2 fields are used in place of the <Property Table References>' index_on_axis fields.

See the examples for the <Image> class.

CONSTRAINTS:

  1. The <Image> shall have at least one <Property Table Reference>. (The data producer may have as many <Property Table References> as desired, as long as there is at least one, and bits_of_material_# will handle them.)
  2. All the <Property Tables> being referred to by the <Property Table References> shall be of the same size, since the material reference corresponds to all of them.
  3. bits_of_material_1 + bits_of_material_2 + bits_of_material_2_percentage shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) shall be zero.
3.2.2.14 THREE_MATERIALS

Used to indicate that each texel in the <Image> represents a linear combination of 3 materials in the <Property Tables> referenced by this <Image>. That is, each texel consists of 4 values: 3 indexes into the <Property Tables> referenced by this <Image>, and the percentages (integers, 0 - 100%) of materials 2 and 3 These <Property Tables> describe the materials. Normally, a <Property Table Reference> is used to find the corresponding <Property Table>, find which <Axis> is referred to, and which hash value measurement along that <Axis> is being referenced. The bits_of_material_1, bits_of_material_2, and bits_of_material_3 fields are used in place of the <Property Table References>' index_on_axis fields.

See the examples for the <Image> class.

CONSTRAINTS:

  1. The <Image> shall have at least one <Property Table Reference>. (The data producer may have as many <Property Table References> as desired, as long as there is at least one, and bits_of_material_# will handle them.)
  2. All the <Property Tables> being referred to by the <Property Table References> shall be of the same size, since the material reference corresponds to all of them.
  3. bits_of_material_1 + bits_of_material_2 + bits_of_material_2_percentage + bits_of_material_3 + bits_of_material_3_percentage shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) shall be zero.
3.2.2.15 IMAGE_INDEX

Used to indicate that the <Image> consists of references to other <Images> (i.e., each texel within the <Image> is the index of another <Image>). This mechanism allows an <Image> to define high-resolution insets.

Each texel is to be replaced by the entire <Image> identified by the <Image> whose index is specified by that texel. This allows data providers to create a gigantic <Image> formed by many smaller <Image>. The index is an index into the <Image Library> containing this <Image>.

FURTHER CONSTRAINTS:

  1. bits_of_image_index shall equal the size per texel (a positive value), while all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx fields) shall be zero.
  2. Each texel within the <Image> shall resolve to a valid index within the <Image Library> containing this image, but neither directly nor via other SE_IMAGSIG_IMAGE_INDEX <Image> instances to the index of this <Image>. (This would cause infinite recursion when trying to resolve the image to its component parts).
  3. All referenced <Image> instances shall have the same values as the main <Image> for colour_model, data_is_little_endian, data_is_3D, component_data_type, scan_direction, and scan_direction_z.
  4. All referenced <Images> shall have either the same image signature X, or SE_IMAGSIG_IMAGE_INDEX which resolves to referenced images with signature X, so that the main <Image> can be resolved to a single image signature.
3.2.2.16 BUMP_MAP_UV

Used to indicate that the <Image> represents a bump map in UV space, specifying information used to modify the surface normals of a smooth surface. See SE_IMAGSIG_BUMP_MAP_HEIGHT for further information on bump maps.

FURTHER CONSTRAINTS:

  1. bits_of_bump_map_u + bits_of_bump_map_v shall equal the size per texel (in bits); all other bits_of_xxx fields (and their corresponding min_value_of_xxx, max_value_of_xxx) fields shall be zero.
  2. When an object has K <Image Mapping Functions> providing its texture mapping information, at most one of the K <Image Mapping Functions> can legally provide a bump map. (Multiple bump maps produce an undefined result.)
3.3 Mipmapping

For a detailed discussion of mipmapping, see [OPENGL], Chapter 9, "Texture Mapping", the section on advanced multiple levels of detail.

3.4 <Image Anchor> and Other Components

An <Image> instance may itself directly represent environmental data, as in the case of raw satellite imagery, for example. In such cases, a <Image> instance may itself require a mapping to a spatial reference frame, independent of any <Image Mapping Function> instances.

An <Image Anchor> instance, when used as a component of an <Image> instance, indicates that the <Image> in question is specific to the given region in the environment. (When the region represents part of the surface of Earth, the image is sometimes said to be geo-specific.) That is, the <Image Anchor> defines a spatial reference frame, three <Location> instances in that spatial reference frame, and a mapping between those <Location> instances and the <Image>, so that the <Image> is "anchored" to that specific region in space.

When an <Image Anchor> is interpreted as a component of an <Image>, the <Location> components of the <Image Anchor> are interpreted as follows.

  1. The first <Location> component specifies the location to which the lower left corner of the <Image> is mapped.

  2. The second <Location> component specifies the location to which the upper left corner of the <Image> is mapped.

  3. The third <Location> component specifies the location to which the upper right corner of the <Image> is mapped.

The three <Location> components of an <Image Anchor> instance are interpreted as the locations of corners of the given <Image> instance, NOT those of the centre of the texels in the corners of the <Image>.

3.5 <Image Mapping Function>

An <Image> that does not have an <Image Anchor> component can be referenced by other SEDRIS objects by means of <Image Mapping Function> instances; see that class for details.

See also the following classes, as well as applicable types and constraints referenced by those classes.



Return to:Top of this Page, Table of Contents

Last updated: July 26, 2006 Copyright © 2006 SEDRIS