AI: The future of remote production is already here
Producers, content generators and commercial brands are under constant pressure to speed up event production, thus meeting the expectations of audiences for instant news and media coverage for any event. At the same time, budgets available for production decrease and innovative production methods are required instead of more traditional deployments, which make excessive use of human and material means and are therefore, as a general rule, expensive.
By Juan Dorrego and David De No
Is then the introduction of AI algorithms in production media the innovation we are looking for?
Fully automating news production without need for camera operators and producers maybe still has a long way to go, but aided production is already here with us. In fact, there are practical examples of automated camera operators, virtual direction and automated production programs in sports such as football. These solutions are already being used by small producers or companies engaged in streaming, but projections are that in the next few years major TV stations and producers will also embrace these new technologies. We are facing the same situation as when DSNGs were replaced by small broadcasting units the size of a backpack.
Therefore, it is important for the industry that AI-based technology is adopted now so we can get acquainted with its capabilities. A clear, recent example of this trend is a Belgian producer that built a news set that did not require human operators and used instead AI systems for camera control and live broadcast. An anchor was only required for emergencies and situations needing immediacy and responsiveness in broadcasts such as news coverage in serious or catastrophic events. Just by having on the studio an anchor or a reporter and at the touch of a button the robot-operated set springs up to life and so does the broadcast.
As for sports events, fully-automated production without need for production staff remains nowadays a distant goal, although AI-aided productions are a very real one.
One of the obstacles for a fully-automated broadcast is the time required by the AI algorithm to learn what is really interesting or important in each sport. For instance, an algorithm may regard that tracking a quarrel during a football match is the same as tracking a punch in a boxing fight, while they are two things completely different for the viewers: the first is an exceptional incident and the second completely normal in the combat.
Achieving an optimal development of algorithms takes time, so the human factor is still very much needed. AI, for example, tends to follow people running or walking rather than people lying on the pitch, so it necessarily does not follow a player falling to the ground after a play, which is what viewers actually want to watch.
Read the whole report in: https://issuu.com/daromedia/docs/tmbroadcastmagazine67/24