ANIM (File Format): Difference between revisions
Jump to navigation
Jump to search
→Rotation Vector Bitstream Format
imported>Jackoalan |
imported>Jackoalan |
||
| Line 135: | Line 135: | ||
[[wikipedia:Delta encoding|delta encoded]] and [[wikipedia:Quantization (signal processing)|quantized]] | [[wikipedia:Delta encoding|delta encoded]] and [[wikipedia:Quantization (signal processing)|quantized]] | ||
to make for a very compact, compressed animation. The data is processed sequentially to reconstruct the | to make for a very compact, compressed animation. The data is processed sequentially to reconstruct the | ||
uncompressed bone channel data. | uncompressed bone channel data as quaternions. | ||
In addition to the bitstream quantization, the format includes a bitmap specifying which frames of the animation have | In addition to the bitstream quantization, the format includes a bitmap specifying which frames of the animation have | ||
| Line 146: | Line 146: | ||
Once resolved, rotation components are divided by ''rotation_divisor'', resulting in an axis-angle | Once resolved, rotation components are divided by ''rotation_divisor'', resulting in an axis-angle | ||
[[wikipedia:Axis–angle_representation#Rotation_vector|rotation vector]]. | [[wikipedia:Axis–angle_representation#Rotation_vector|rotation vector]]. This rotation vector is converted | ||
to quaternion imaginary components with the W component derived via normalization and a ''sign-bit'' | |||
packed into the bitstream: | |||
<pre> | |||
q = π / 2.0 / rotation_divisor | |||
X = sin(rx * q) | |||
Y = sin(ry * q) | |||
Z = sin(rz * q) | |||
W = sqrt(MAX((1.0 - (X^2 + Y^2 + Z^2)), 0.0)) | |||
W = sign_bit ? -W : W | |||
</pre> | |||
Translation components are multiplied by ''translation_multiplier'', yielding world units. | Translation components are multiplied by ''translation_multiplier'', yielding world units. | ||
| Line 189: | Line 199: | ||
|Word-packed bitstream laid out as described by the descriptor table. The bitstream is unpacked just like ''key_bitmap'' (least-significant to most-significant). | |Word-packed bitstream laid out as described by the descriptor table. The bitstream is unpacked just like ''key_bitmap'' (least-significant to most-significant). | ||
Key values are tightly-packed, quantized integers whose bit-depth is specified by the ''q_*'' values in the descriptor table. | Key values are tightly-packed, quantized integers whose bit-depth is specified by the ''q_*'' values in the descriptor table. | ||
Each bone-channel key-set starts with a single bit providing the '' | Each bone-channel key-set starts with a single bit providing the ''W sign-bit'' (1 for negative, 0 for positive). | ||
The key integers are signed using [[wikipedia:Two's complement|Two's complement]] representation. | The key integers are signed using [[wikipedia:Two's complement|Two's complement]] representation. | ||