Week 5 Video on the Internet

2 Overview Video & Internet: The problem Solutions & Technologies in use Video Compression Available products Future Direction

3 Introduction World Wide Web is no longer silent Technologies brought sound, video and animation instead of textual HTML pages and speechless still images Difficulty begins when audio/video file is transmitted while it is being created, and converted at the receiving point into a continuous video and sound. Demand can be real time, i.e. small delay between sender and receiver

4 Video & Internet: The problem

5 Video Bandwidth needs When trying to transfer live video, the bandwidth problem becomes much more critical. To transfer a video picture (captured at a low resolution video camera) - a compression ratio between 2500 to 5000 is needed. Increasing the bandwidth is possible in several ways, all of them are expensive: –Cable modem which will be connected to the cable network and will use the high Coax cables bandwidth. –Fibre Optics cables which in labs achieves bandwidth of billions bits per second - theoretically far from exploiting this potential.

6 Internet Delays Internet uses telephone system - its main strength is the ability to transmit real-time continuous speech. Unreliable when it comes to transmitting data. Internet's main strength is it's ability to transmit asynchronous data - unreliable if the data transmission is required in real time because: –Heavy traffic load and –Internal transmission problems can cause delays This results in disrupted speech and video reproduction at the destination computer. Disruption in an Internet transmission is purely a gap in the data stream reproduction.

7 Video compression Compression of the video signal is done in a number of stages, which causes loss in the visual information: –Video camera captures only a small part of the view area in a low resolution. –Conversion of analogue signal that comes from the camera to a digital one in 18 Mbps. –Conversion of the 3D colours of RGB into 2D presentation –Reduction of bit rate to lower than 4 Mbps in a "brutal" way before the smart compression.

8 Compression techniques Compression ratio can be achieved in a couple of ways: Looking at a pattern of video frames as a "3D" information. Two dimensions of horizontal and vertical and the third dimension is the time. Compression in the time dimension is different than the static "2D" image, because the time pattern needs to transmit only the changes from frame to frame. By using the base frame it is possible to make continuous frames by transferring small amount of data that describes the differences between two continuous frames, until a new base frame should be transferred.

9 MPEG compression Technique that can be used is performing interpolation to guess how the pattern is created from the beginning to the end. MPEG compression is based on this method –three kind of frames called I-frame (intra frame) P-frame (predicted frame) B-frame (bi-directional interpolated frame). 2D compression JPEG of the static frame is used to compress the base frame (I-frame). The JPEG algorithm is based on spectral analysis of the frame and focusing on the major frequency components of the image. The required image quality sets the compression ratio. Changing the compression in the time dimension allows setting the number of interpolated frames (B-Frame) and the predicted frames (P-Frame) that are stored between one base frame to another.

10 Wavelet compression Compression technology today, allows transferring a continuous video in 15 frames per second and a resolution of 320*240 pixels on a digital line of 256 Kbps. To achieve a similar quality under standard telephone lines a 10:1 compression ratio is required. One of the technologies today that might deliver this compression ratio is called WaveLet. One of the most attractive aspects of the WaveLet compression is the ability to make a degraded improvement. The same compression creates different "layers" of details, with different quality, so the quality can be improve by combining more and more layers in the frame, when the bandwidth allows it.

11 Available products Two products currently support streaming video technology: –Stream Works, from Xing Technology corp. –VDOLive from VDOnet corp. VDOLive uses a proprietary algorithm based on Wavelet compression Stream Works which uses a derivative of MPEG-1 put emphasis on the quality of a static picture.

12 Streaming - The Concept Download an initial portion of the video or sound to fill a local buffer Start playing the video or sound while the source continues to load the buffer Result: visual or audible results shortly after clicking on a link Internet Buffer 

13 Streaming - Pros and Cons Almost immediate results after clicking on link Can degrade result while communications slow Offers a degree of copy protection  Cannot guarantee quality of viewed result  Difficult to save final result for repeated views  Delay if you want to playback non-sequentially  Consumer (usually) needs to install viewer

14 Typical Products RealAudio RealVideo RealPlayer G2 –Basic version free, Plus version $30 Server requires RealServer, costing $2000+ Trend towards Java applet viewers Merging with video on demand

15 Future Direction As the Internet grows in global size and bandwidth, and as computer technology increases in speed and drops in price, Internet Video Conferencing will become increasingly more feasible. Ultra-efficient CODEC software, full duplex sound boards, and fast Internet connections will soon bring low-cost, CD- quality telephony and high quality video to all Internet users. The next generation of Internet Video Conferencing will also include a number of interesting features such as voice mail and on-the-fly data encryption. We are at the very beginning of a revolution in communications, and Internet Video is the ultimate first step.