Producing Content for Mobile Delivery
A version of this article will appear in the 2008 Streaming Media Industry Sourcebook, which comes out in February 2008. It's free to all Streaming Media magazine subscribers—get your free subscription here.
Wireless video delivery is complicated. Not only is different technology deployed from market to market, but the nature of radio frequency (RF) dictates that bandwidth and throughput conditions change dynamically. Video-enabled handsets are available in all markets, but the capabilities of the handsets differ widely. Many different delivery technologies also are in play, including GPRS, EDGE, EVDO, UMTS, HSDPA, DVB-H, MediaFLO, Wi-Fi, and WiMAX. As 3G and post-3G networks are deployed, more bandwidth will be available to video, which will increase delivery options. But the bandwidth within a video session will still be variable, and as video adoption increases, scarcity of spectrum issues will remain.
Producing video content for wireless networks can be as complicated as delivering it successfully. Content comes from a variety of sources, and it must be encoded so that the video will stream easily within the network and will play successfully on the mobile handset.
Some content providers produce video specifically for mobile delivery while others repurpose content produced for broadcast media. Still more content is produced by end consumers on-the-fly. To make all this video work effectively, we need a new understanding of production processes, driven by the realities of the wireless delivery channel. There are three objectives that should be kept in mind:
1. Create a viable product that will be purchased more than once.
2. Provide a positive user experience for a broad audience.
3. Meet or exceed the expectations of this new audience.
In order to accomplish these goals, one must understand the intricacies of video compression in order to tailor production in a way that optimizes the output from the viewer’s perspective.
Understanding Codecs
Mobile video must be squeezed into the available wireless network channel. The act of squeezing (compression) can produce unsightly artifacts, adversely affecting the viewer’s quality of experience and, thus, the perception of the content provider’s brand. Preprocessing video and audio assets prior to encoding can have a dramatic affect on minimizing the artifacts.
A codec is a COmpressor and DECompressor of digital video. Encoding is the use of codecs to reduce the size of a digital video by removing redundant visual information. Encoding can produce visual artifacts since it is a "lossy" process.
Wireless and mobile video are bitrate-limited delivery channels. Best practices dictate that the content must be matched to the codec’s needs for mobile delivery; i.e., it should not exceed the available bitrate in the mobile viewer’s channel.
Codecs use a number of techniques to compress the size of digital video content:
• Spatial Compression: Each pixel within a frame is compared to neighboring pixels. Based on a given threshold, pixels are evaluated as similar or different and are discarded if considered redundant.
• Temporal Compression: Successive frames of video are compared to one another, and a series of keyframes (I frames)—containing all the data that comprises the image—are created to establish a baseline quality. Delta frames (P and B frames) fall between I frames and are more highly compressed, holding only the pixel information that has changed from the previous frame and requiring a decoder to reference other frames to present a complete image.
• Color Space Subsampling: This process exploits the fact that human eyes are more sensitive to brightness than to color, so it reduces the amount of color information.
• One-Pass: One-pass encoding is speed-optimized, whereby the codec analyzes the need for bitrate slightly ahead of the actual encoding. It is used in applications requiring real-time encoding, such as live streaming.
• Two-Pass: Primarily used in nonreal-time applications such as video-on-demand, two-pass encoding analyzes the bitrate requirement in a complete pass through the video, and then it applies the bitrate map to the video in a second pass.
• Constant Bitrate (CBR): Video is encoded at a bitrate below a ceiling, usually specified by the mobile network operator. Video bitrates above the ceiling would cause interruptions in delivery.
• Variable Bitrate (VBR): Bitrate varies with demand of the content, producing higher-quality output. Dynamic, high-motion scenes require more bitrate than talking heads. Spikes in encoded video bitrate are acceptable depending on the delivery medium because they provide higher quality, but they don’t work well in narrowband networks.