Offline motion description for video generation

A method that when executed by a computing device extracts motion data, the method comprising:

• obtaining motion description information for a macroblock in a frame of multimedia;
• calculating partition modes using a Lagrange cost function based on a predicted residue that corresponds to one or more encoded content features and a motion residue of a motion vector in the descriptive motion information;
• selecting an optimal mode from the calculated partition modes using the Lagrange cost function;
• and
• selecting an optimal motion vector that corresponds to the selected optimal mode using the Lagrange cost function.

A computing device, comprising:

• a processor;
• a memory into which a plurality of instructions are loaded, wherein upon execution of the plurality of instructions by the processor, the computing device performs acts comprising:
• organizing motion data in a plurality of macroblocks of an image based on a hierarchical model that includes a plurality of partition modes of various sub-unit sizes, the plurality of partition modes including a partition mode that is converted from a skip mode and a direct mode in order to reduce dependency between a macroblock and at least one other macroblock;
• coding motion information of the image offline based on correlations between spatially neighboring macroblocks that includes motion vectors to form a motion description;
• identify a Lagrange multiplier within the motion description created offline that corresponds to a quantization parameter;
• select a mode that corresponds directly to the Lagrange multiplier as an optimal mode;
• and
• select an optimal set of motion vectors from a plurality of motion vectors of the optimal mode using a Lagrange cost function.