Each light event (flash) is renderized independently. The final image is obtained by summing up the partial results weighted by duration of each light event.
To identify the magnitude of the object's movement in the image, the World Bounding Box of the object is projected over the film. A pair (Time (t), Image Bounding Box at t) is called an Instant Descriptor. The Instants Descriptor are the features used to refine the movement of the object over the image.
Given a certain light event, we find the Instants Descriptors associated to t=0, t=0.5 and t=1 of the event (those times not necessarily coincides with the initial, mid or final exposure times of the camera). If the Image Bounding Box (IBB) at t=0.5 is farther to IBB at time t=0 than to IBB at t=1, then we find an Instant Descriptor at t=0.25 to refine the movement. Otherwise, we find an Instant Descriptor at t=0.75. By building a heap structure we can easily get what are the consecutive Instants Descriptors whose IBB are the most separated and keep refining the object movement. The metric used to define the distance between IBB's is the sum of the squared lenghts of their respective Lowmost-Leftmost corners and Uppermost-Rightmost corners.