Challenge

Product placement has become a familiar feature in entertainment. As conventional advertising finds itself under threat from set-top boxes that allow users to edit out commercial breaks, the opportunity to digitally place brands within film and TV content is becoming increasingly valuable. Special effects technology means product placement is no longer a matter of ensuring an actor is wearing the right watch or using the right phone during filming. Branding can be digitally embedded within video to appear part of the scene, providing companies with the added value of localised, tailored advertising that can’t be skipped.

In addition, manually embedding advertising images within videos is a labour-intensive process requiring hours of painstaking work by graphics professionals. Not only must the image be positioned correctly but it must react to changes such as camera angle, lighting and objects passing in front of the image to appear as a seamless addition. The high cost associated with this process has limited the revenue potential for embedded advertising.

Solution

UK-based embedded advertising specialist MirriAd has developed a range of patented solutions that provide a cost effective method for embedding passive or interactive advertising (video, image or text based) directly into existing video assets. The automated embedded advertising solutions use NVIDIA graphics processing units (GPUs) and dramatically accelerate the process by analysing videos to identify suitable advertising zones. Once these zones are pinpointed, images can be inserted and will react to movement and tonal changes within the frame as if they were part of the original scene.

Although this process can take place at any time, the brand references appear as if they were incorporated in the original production. Those brand references may include physical objects such as vehicles, or, for instance, still or video images on a billboard. The entire process dramatically simplifies what was previously a highly complex and expensive process when carried out post production. The advertising is unobtrusive to the viewer since they see it as part of the original, and it can’t be skipped.

MirriAd’s solutions mean that content owners can provide advertisers with the means to address an increasingly fragmented market and to take advantage of that fragmentation by inserting advertising, specific to the market variables. Although essentially the same content may be available on a wide variety of platforms and devices, and in a range of geographies, brand owners can insert market-specific advertising for each of those outlets.

MirriAd has used NVIDIA graphics processing unit (GPU) processing technology to achieve impressive speed-ups over conventional CPUs, including 30 to 100 times faster rendering speeds. The company’s success with NVIDIA GPUs prompted their decision in 2007 to develop exclusively with NVIDIA technology.

Impact

MirriAd already counts major film and TV organisations worldwide, including Comcast and ABC Family among its customers. The company is now looking beyond its core broadcast business to new markets in security, defence and online video placement. Its work was recently recognised with a £1.5m grant from the UK Government as part of its Creative Industries Innovation Award programme to develop VIDEOSENSE™, the world’s first tool for intelligent automatic video analysis.

Dr Janko Flogel, Chief Technical Officer at MirriAd, explains: “The problem we’re solving with our video analysis application is very well suited to the GPU, which is why adopting NVIDIA’s technology has been so beneficial. The speed-ups we’ve already achieved using GPU computing are very exciting and we’re now porting other parts of our application, including video decoding and motion detection algorithms, to take advantage of its parallel architecture.”

He continues: “We are also building a server cluster based on NVIDIA Tesla 1U Computing Systems. This will become a core part of our VIDEOSENSE system, enabling users to manage the image placement process themselves. In an environment where users have a very low tolerance for online advertising in the form of banners and pop-ups, VIDEOSENSE has enormous potential for monetising online video in a way which is seamless for consumers and extremely valuable for advertisers. This facility is only possible thanks to the cost-effective processing power provided by NVIDIA Tesla.”