5.2. Rendering computationally demanding
visuals
We anticipate that artists will at times desire visuals
whose computational complexity exceeds that which can
be rendered in real-time with the available processing
power. To cope with this eventuality, we provide a simple
mechanism that allows Sonnet+Imager to render complex
visuals in an off-line (non-real-time) mode, for later
playback.
Specifically, the Imager virtual clock enters a mode in
which it only advances time when the current frame is
finished rendering, no matter how long that takes.
Likewise, the Imager engine enters a mode in which it
records each frame (once fully rendered), along with its
virtual timestamp (typically a single frame’s worth of
time). The resulting performance is a sequence of frames,
suitable for compression by any of the various available
techniques (e.g. QuickTime, MPEG). Note that no other
portion of the architecture needs modification to support
this mechanism.
With this mechanism in place, the artist gains access to
a larger space of visuals, at the expense of the ability to
see the performance in real-time. Obviously, this makes
developing complex performances harder. To compensate