There has always been a problem of compatibility between videos and screen sizes, further compounded by resolution and aspect ratio differences. Plus there is always the problem of running videos on mobile devices. But it looks like these problems will soon be a thing of the past, thanks to Google Research which has just announced an algorithm for resizing videos without compromising on quality and viewing pleasure.
This newly devised algorithm will retain the width to height ratio which is important for preserving the original video quality and size. The algorithm also distinguishes between salient and non-salient parts of a video. Now you may wonder what exactly is this all about.
Well, according to the Google Research team, salient content refers to indispensable elements on the screen such as people and objects which are in primary focus. These are important parts of a video which a viewer expects to be unaltered no matter how the video is resized. Non-salient elements are objects/people in the background and even the background scenery. The algorithm stretches or reduces these elements without causing a change in the overall look of the video.
This technology was devised with the help of Georgia Tech researchers and is called discontinuous seam carving because of the way it divides salient and non-salient parts of a video. At the same time, the vital aspect ratio is for each important part of the frame is retained and distortion is minimized for the non-important parts. So, the overall change in the video after the resize is not too radical.
Videos are processed by this algorithm one frame at a time, and so, streaming video can be resized for proper display on various mobile devices and PC screens, using this technology.
Resizing Videos on YouTube?
Though Google has no plans to include this in YouTube, this algorithm is an amazing prospect. The Research team identifies a few shortcomings such as the lack of precision in recognizing salient parts of a frame. An interface has been provided which will allow you to highlight the salient parts of the frame on your own. But further improvements will be needed because this can come in the way of your viewing experience.
Moreover the video resizing algorithm fails to deliver in fast sequences or frames where there is an excess or a lack of salient content. So, it is safe to say that the algorithm is still far from the finished product.
