HP ProLiant 1600 Video Streaming Technology - Page 11
An I-frame of the GOP
View all HP ProLiant 1600 manuals
Add to My Manuals
Save this manual to your list of manuals |
Page 11 highlights
ECG068/0798 WHITE PAPER (cont.) 1...1 Depending on the amount of motion in the video, P-frames may not come fast enough to give the perception of smooth motion. To compensate for this, B-frames are therefore inserted in between the I and P-frames. B-frames are "bi-directional" in that they use information in the previous I or P-frame as well as in the future I or P-frame. The information they contain is interpolated between two endpoints rather than representing actual data, making the assumption that the pixel information will not change drastically between the end-points. As a result these "dishonest" Bframes contain the most amount of compression and are the smallest frames. In order for a decoder to decode these B-frames it must have the corresponding I and P-frames they are based on, hence the frames may be transmitted out of order to reduce decoding delays. A frame sequence consisting of an I-frame and its subsequent B and P-frames before the next Iframe-typically around 15 frames-is called a Group of Pictures (GOP). An I-frame of the GOP acts as the basic entry access point. The result of this compression technique is that MPEG format is not easy to edit since you cannot enter the video at any point. Also the quality of the resulting video will depend on the particular implementation and the amount of motion in the video. The MPEG encoding/decoding process can also be seen as computationally intensive which requires the use of specialized hardware or a PC with a powerful processor. The compression achieved using this technique enables about 72 minutes of video on a single CDROM-not quite enough for a full-length feature movie, which typically needs about 2 hours or two CD-ROMs. In addition to compression techniques, the MPEG-1 standard also supports playback functions such as fast forward, fast reverse, and random access into the bitstream. However, as mentioned above, these access points are only at the I-frame boundaries. The CDROM quality audio is stereo (2 channel) 16-bit sampled audio at 44KHz. The main MPEG-1 resolution of 360 x 242 is called Standard, or Source, Input Format (SIF) and, unlike the Common Intermediate Format (CIF) defined earlier, differs for NTSC and PAL systems. The computer industry has defined its own square pixel version of the SIF format at 320 x 240. Originally, MPEG-1 decoding was done by hardware add-in boards with dedicated MPEG audio and video decoder chips because of the processing power required. Currently, MPEG-1 can be decoded in software on a Pentium 133 or better. MPEG-2 MPEG-2 was adopted in the Spring of 1994 and is designed to be backward compatible with MPEG-1. It is not designed to replace MPEG-1, but to enhance it as a broadcast studio-quality standard for HDTV, cable television, and broadcast satellite transmission. Resolution is fullscreen (ranging from NTSC 720x480 to HDTV 1280x720) playback with a scan rate of 60 fields per second. The latter enables MPEG-2 to support interlaced TV scanning systems as well as progressive scan computer monitors. Compression is similar to MPEG-1, but is slightly improved with the video exceeding SVHS quality, versus VHS for MPEG-1. Audio is also improved with six channel surround-sound versus CD-ROM 2-channel stereo audio for MPEG-1. The use of Interframe compression similar to MPEG-1 makes fast-motion scenes the most difficult to compress. Because of this, MPEG-2 supports two encoding schemes depending on the needs of the application: variable bit rate, to keep the video quality constant, and varying quality to keep the bit rate constant. The variable data rates for MPEG-2 range from 2Mbps to 10Mbps. This typically requires a 4x CD-ROM drive for playback. A 4x CD-ROM can only store only 18 minutes of video versus a 1x CD with 72 minutes of video for MPEG-1. MPEG-2 was selected as the core compression