Analysis of properties of effects for rendering and caching of media data

Abstract

Performance of rendering of effects on media data may be improved by analyzing properties defined for an effect to identify static and dynamic time segments within the effect. A static time segment in an effect is a time segment for which only one image is rendered. A dynamic time segment in an effect is a time segment for which each sample of media data in the time segment is rendered. An effect can contain any number of static and dynamic time segments. By identifying these time segments in an effect, processing and caching of the rendered effect are more finely grained. In particular, rendered time segments in an effect may be cached, such that a change in the effect in one time segment does not require re-rendering of other time segments. Thus, unnecessary rendering may be avoided, and invalidation of cached rendered media data and subsequent rerendering also may be reduced. In an animated effect, only one sample of media data may be rendered for a time segment in an effect if property animation and input material are both static in the time segment. If a key frame is inserted into a function curve of a property for an effect in such a way that the curve does not change, then any cached media data previously rendered for the effect need not be invalidated for the effect or for a time segment in the effect that contains this key frame. For a time segment in an effect in which properties do not change input media data in any fashion, no rendering is needed. Such a time segment may be either a static or dynamic time segment, depending on the input to the effect.