An apparatus, system, and method utilizing multi-description coding (MDC) for multimedia content with a plurality of forms of diversity. A source scene can be fragmented into one or more region of interest (ROI) portions and non-ROI portions, based on motion or any other metric of interest. One or more of these portions can further be fragmented by sub-sampling the respective portions to generate a plurality of lower-resolution versions, e.g., with alternating groups of pixels in respective versions. Still further, one or more of these portions can be further fragmented by image frames, e.g., with alternating frames in respective fragments. At least one ROI portion and lower-resolution versions may then be encoded into a plurality of descriptions and transmitted. Utilizing various combinations of ROI diversity, resolution diversity, and frame rate diversity can improve channel utilization and robustness for streaming multimedia content.