For optimization, flexibility and rendering speed, the frames of your project are divided into parts (tascas).
Each task is rendered in parallel by a separate server (currently free). As a consequence, the output frames may appear in different sequences.
Since the scene is rendered on several servers, it is necessary to cache the parameters, patikls, etc. You can read more about this in the "Preparing a project" section.
The rendering of a single task consists of several stages indicated at the diagram:
- scene opening: the larger the scene file size is, the longer it is
- preparation: texture conversion, parametric stuff miscalculation, geometry unpacking, etc.
- rendering directly by the videocard core: it depends on the number of passes, re-reflections, SSS, volumlights, etc.
- saving sequences: the more resolution and more passes - the longer
Frames in the project have different complexity, which can also increase or decrease render time. This time can not be predicted and put in the calculation, so the calculator's forecast and the actual rendering time can differ significantly.