File Size Reduction
Conventional compression technology and existing network architectures are challenged to scale adequately to meet the rising demand for high quality video. Internet usage was originally dominated by many small transmissions of text, and eventually images all of which are much smaller in size than video data. The problem of the disparity of these data types then gets worse. As the world upgrades devices, video resolutions move from small format to standard definition, thenhigh definition and beyond, already large videos continue to grow in size.
The eFLEX codec prototype delivers greater video compression rates than conventional compression technology without sacrificing visual quality. The resulting file size and data stream size reductions provide another means to relieve bandwidth, latency, and storage constraints inherent in modern computing infrastructure.
Downloading a 150MB movie trailer, takes about five minutes using a reliable 4Mbps broadband connection. Downloading a 4GB high-definition movie takes over two hours to complete over the same connection. As most networks are asymmetrical, uploading a file takes significantly longer. Technologies like buffered streaming and progressive downloading mitigate part of the problem because they allow people to watch a video without first downloading the entire file. These technologies, however, do not directly address the need for reducing the whole video.
As people continue to demand higher quality video, one can see the effect on the video’s file size by considering one frame of uncompressed Internet standard video grows in size by 27 times with high definition resolution:
.jpg)
Advances in the areas of Computer Vision, Image Understanding and Pattern Recognition provide ever more effective ways of analyzing and modeling phenomena in video data. At higher resolutions the efficiency of conventional compression technology typically levels off while the opportunities for empirically derived models increase. eFLEX is built with EuclidVision, a suite of tools, algorithms and processes that draws on these advanced algorithms and modeling techniques to provide an alternate means to reach higher video compression without a loss in visual quality.
Much of eFLEX is backward compatible with MPEG-4 Part 10/H.264 in terms of using the convention as a basis and then building the higher level modeling using the underlying H.264 algorithmic elements. Additional enhancements made to H.264 encoders in additional to new standards build on extensions of the H.264 standard are complimentary to eFlex and do not conflict with the high level processing techniques used by EuclidVision and the eFLEX codec, enabling implementation within today’s networks.