When all pixels of an image are considered to be completed, then the rendering task is considered to be complete. Between the rendering engines and the merge process, the merge process may be better able to make this determination because the merge process has better knowledge of all of the rendering results.

When the merge process determines that the merged rendering results meet the halting condition, the merge process instructs the rendering engines to halt the processing of the pixels of the image. For example, the merge process may send a “completed” order to all of the rendering engines (see, e.g., dotted-line arrows leading from merger 140 to rendering engines 130 in FIG. 1). Each rendering engine stops execution upon receiving such an order.

In the area of multi-machine rendering, according to one approach, tasks are dispatched based on image-area-based subdivision. For example, each rendering engine is assigned a sub-region(s) of the image and makes the halting determination for that sub-region(s) independently. Because a particular rendering engine computes its entire sub-region independently, the rendering process has knowledge of the information that is required to make the halting determination for that sub-region. However, before computing the sub-region, the rendering engine is unable to know how computationally complex the sub-region will be.