Graphics Reference
In-Depth Information
To implement buffers, glTF borrowed two fundamental types from typed ar-
rays:
•
ArrayBuffer
for entries in buffers property, and
•
ArrayBufferView
for entries in the
bufferView
s property.
Both
buffer
sand
bufferView
s implement the same interface of their respective
typed-array counterpart. However, types are not named exactly the same because
glTF
buffer
sand
bufferView
s contain additional properties. In the JSON snip-
pets in Figure 2.4, properties in blue come directly from typed arrays. Properties
in red are not part of typed arrays and were added to provide
•
external storage via a path property,
•
a hint about the kind of data stored via a target property, which is optional
but a useful optimization for binding the corresponding VBO once and just
updating the offset for subsequent draw calls.
”buffers”
:
{
”office”
:
{
”byteLength”
: 829958,
”path”
:
”office .bin”
}
}
,
”bufferViews”
:
{
”bufferView 7958”
:
{
”buffer”
:
”office”
,
”byteLength”
: 686264,
”byteOffset”
:0,
”target”
:
”ARRAY BUFFER”
}
,
”bufferView 7959”
:
{
”buffer”
:
”office”
,
”byteLength”
: 143694,
”byteOffset”
: 686264,
”target”
:
”ELEMENT ARRAY BUFFER”
}
}
While buffers and
bufferView
s provide access to data, other concepts are
needed:
•
Types:
In order to map directly to GL APIs, glTF reuses existing GL types.
•
Interleaving:
Typed arrays deal with contiguous buffers, but to interleave
vertex attributes, other properties like stride are needed. Also, semantically,
count fits better than length.
