Cracking the Video Encoding Code

Big changes are afoot in the broadcast business and a fault-line is appearing in the industry as content distribution pushes further online. Major announcements from large media companies backing online video channels and pushing premium 'TV' content out without digital rights management over the Internet are shaking up the industry. Where this is leading us is a question on many people's lips as we stand either side of this digital divide wondering what innovation is going to fill in the gap. One thing is for certain, whether program content is delivered online or through more traditional broadcast channels, the issue of image quality remains as important now as ever. Online video is made possible through the use of video compression technologies and broadband Internet. They have become much more efficient in recent years, allowing us to squeeze video of reasonable quality to audiences around the world at rates that would have been impossible a few years ago. However, the process of encoding video, one could argue, has become oversimplified of late and this has resulted in a lower than achievable quality of content being pushed over the Web. Every day we see video appearing on high profile networks, professionally produced content that has not been correctly encoded for online delivery. It looks bad, but it does not have to be this way! An unnecessarily high level of encoding artifacts are present in many video items online that result from the transcoding of video from one format to another at different bitrates. It would also seem that a large percentage of online video is being pushed out as interlaced video. This is not broadcasting and the audience is not seated in front of television sets. The online audience consumes their content through iPods, LCDs, PSPs and a myriad of other devices that are all designed for progressive scan video. The upshot of all of this is low quality video, a sub sub-optimum viewing experience and fatigued audiences. As with the business of TV, keeping the audience engaged is the key to staying in clover, and delivering low quality images can cause a viewer to go elsewhere, perhaps another platform or perhaps ditch the content altogether. Delivering the best quality video within the bandwidth constraints that exist will mitigate this risk and serve to maintain the interest of your online audience. How one goes about encoding video for online delivery has a major impact on the quality of the end product and of the user experience. The process should not be simply limited to video compression. There are many small items which when combined and addressed effectively will make your content shine online. At Moving Media's Digital Video Lab, we have learned through our experience, gained through encoding many thousands of video items on behalf of our clients, that building a process that covers a range of key areas ensures a high quality product. Three Steps The following 3 areas are important to address in the encoding of video: Find the best source available The quality of any encoded video file is relative to the quality of the source from which it has been encoded. Any degeneration introduced through the application of compression will remain visible through any further lifecycles of the content. It is important to remember that DVD is already a compressed digital media, ripping files off DVDs. Getting back to the highest quality version as your source is a key objective if you wish to get the best quality. Transcoding existing compressed video files also presents significant quality issues, providing an already compressed digital file as the source to create files of a different format or bitrate will result in further degeneration of the signal quality. All codecs (compression-decompression algorithms) compress video data in different ways. When you have multiple encoding cycles you aggregate the negative effects of all of the codecs used. This issue is seen again and again online with video file being flipped from one format to another depending on the platform, the result is video that is blocky and lacking clarity being delivered at the same expense of video that could look pretty good had it been handled well. Create a High Resolution Digital Master A common phase these days is mezzanine file. It refers to a 'clean' digital master which is used to create copies for streaming or download. The mezzanine file is typically a high bitrate digital file using a production codec (for eg.., MPEG 50i or uncompressed AVI). This file is then used to create proxy files ? repurposed versions of the master file created with specific technical settings ? usually created through an automated or batched process. When creating the mezzanine file a number of key steps must be taken, such as removing slates and breaks from a broadcast master tape, cropping the fame to exclude edge noise that may be present. Cropping can also be applied to exclude any misalignment that may be present in the edge of the frame as a result of the content using different sources with different screen dimensions being edited together. Once this master mezzanine file has been created, you are ready to encode for online video. During the process of encoding, tests should be carried out to check the effectiveness of the de-interlacing algorithms used and the efficiency of the video compression applied. Where possible, it is often worth applying two-pass encoding to the process. Two-pass encoding is a process whereby a video file is analyzed in one pass before the second encode pass. The analysis pass allows the encoding system to more efficiently apply compression to the file which results in better quality video overall. This does take additional time as the encoding system will have to do twice the work but the results are significant at the sort of bitrates we are using for many online platforms these days, for example 700kbps. Don't Forget Audio Audio is an area often overlooked when dealing with online video but it is hugely important. Some subjective quality tests we have done show that an audience can rate a video to be of lower visual quality when in fact the difference in quality between the two test files was only relative to the audio. When browsing video online one will find a wide variation in the audio levels of different video files?some loud, others quiet. This requires us to constantly adjust the volume setting of our computer. This is an important consideration for companies who are publishing a lot of video online that may be from a range of suppliers. As there are no standards for the industry, one must try to balance these issues into the technical plan. Normalization of audio levels is the term applied to creating a consistency of volume for video content, another step in a well-designed encoding process. These three areas should provide an introduction into the areas of consideration when driving for quality of online video, whether you encode video yourself or use a third party to do it for you. It is also worth noting that different encoding systems, although they can create the same file outputs, such as MP4, do so at different levels of quality. The same can be said for de-interlacing systems. We often see great differences between a file encoded through a standard editing systems and a file encoded using a dedicated video encoding system.Simon Factor is CEO of Moving Media Digital Video Labs, a company that specializes in the digitisation and encoding of audiovisual materials. Simon has been involved in the technical side of both broadcast and Internet businesses since the mid 1990s and has witnessed the changes as these two markets converge. After setting up Moving Media in 2003, Simon and his team at Moving Media have delivered a number of high profile projects and are retained by a range of leading companies as technical consultants and encoding partners.

Cracking the Video Encoding Code

Featured Stories