In the Internet era, any piece of digital/digitalised art (be it image, video, audio, 3D etc.) can be anytime and anywhere replicated with a simple click, thus frustrating the artists/producers from a large part of their economic benefits. Imposing itself as a viable solution to the copyright enhancement problem, watermarking represents a research field which exploded in the last decade. The present paper reports on an original video watermarking method for very low rate video based on: (1) the synergy between spread spectrum and informed embedding approaches (patent pending) and (2) an accurate statistical modelling of some real life attacks. The detection is oblivious (it does not require the unmarked object). Firm results concerning transparency (no visible differences between the marked and the unmarked video) and robustness (with respect to both mundane transforms, as the compressions, and malicious attacks, as the StirMark attack) are obtained. The data payload is increased more than 10 times with respect to the state of the art. Beyond traditional watermarking, our method can be applied for emerging enriched multimedia applications: interactive television, video on demand, scalable enriched content streaming, and adaptive indexing.