In many ways, the budding netvideo industry of the '90s resembles the budding video industry of the '70s. The wide variety of compression/decompression (codec) algorhytms, incorporated into a growing assortment of compression products, fed through various server systems and played on a collection of incompatible browser plug-ins makes the ancient war between VHS and Betamax seem Peloponesian.

Generally, the netvideo technology involves four steps; acquiring an edited audio/video file, converting the acquisition to a compressed file, serving the compressed files to one or more streams of output and playing the files on a client system. At present, all manufacturers offer the player software free of charge, usually by downloading it from their websites. In many cases, the server and player software is also free as manufacturers use the Netscape business model (orignally developed in the '70s by drug pushers) of giving the first high for free.

Network video streaming takes an old technology - broadcast television - and applies a new distribution path - interactive, internetworked computers. Although some added benefits may be discovered within the computer distribution model, one might safely assume that the total available market for streaming video might be defined as a fairly well contained subset of the total broadcast and cable television market. Indeed, as the US transitions to digital format consumer television in the next three to five years, the logic of joining computer technology with home entertainment becomes faultless.

Will the unique capabilities of a compute, however, add benefits that will result in a greater market than is currently possible with broadcast and cable television? Consider the fact that interactive video has been around since Apple IIe got connected to a comsumer Betamax deck, and never got off the ground even when Panasonic offered laser disk players that cut the head to tail rewind time to about 2 seconds.

Will the globalization of the Internet - obsolescing the effect of national borders on broadcast and cable franchises - provide some unfortold synergy as more of the global polulation wires in? Will new markets develop in the net that did not develop for TV? Perhaps, but language and culture tended to self-limit the broadcast market when home satelite receivers proliferated prior to the encoding of most Englsih language entertainment channels. Does anyone bother to tune into VREMYA's unencoded satelites to watch Russian TV? Nyet, spasiba.

So where is the money come from to fuel a technological transition that offers, at best, equal quality to a thousand or so on-demand viewers? Who, will prefer video on a net to video on a cable or antenna and what will they watch?

While the future possibilities of at-home, video on demand more than justifies today's rapid ramp-up, my guess is that today the viability of this technology derives from three sources; novelty, marketing and education. The novelty of video on the web, from pornography applications to factoid retrieval, is undeniable. Seeing even a droll marketing video play back, in any resolution, is like seeing a member of the opposite sex naked for the first time. Quality really doesn't matter, it's the fact that the newness that captures attention. Unfortunately, novelty does not last long enough to support an IPO much less a good balance sheet as the rapid shakeout of seach engine companies will attest.

Dismissing novelty then, will marketing and education provide enough user pull to justify long term development of video streaming. I think so. The very existence of the web, regardless of the current difficulty in generating cash flows from a website, bears sufficient testimony to the proposition of reaching a global audience for very little cost-per-thousand (CPM) and video closes the last gap of media services of the net - the most important gap for marketing and education which are demonstration-intensive activities.

An intelligent analysis of this technology, therefore, must include an evaluation of the medium and its costs with a keen eye on the most probably avenues of application. How important is image resolution and smooth playback to the two grown applications? Will one be better equipped to conduct a global training session if the video is clearer or larger? No doubt, but will the cost justify the improvement? Will a product or service sell faster if it is featured in video on a website. No doubt, but will the shortcomings of the medium create a negative impression of the product? Consider these bottom line questions as carefully as you consider the technological results of this review. They are as at least as important and will significanly impact your success in this market.

In addition to the technology and econonmic considerations of the video streaming market, one must also consider the issue of creating video content that compells viewership. Any exhaustive study of existing video streaming websites today will reveal a painful fact; not much of it is worth watching in terms of content or quality.

The most obvious fault in content is the plain fact that most of the video we found was obviously recycled from some other purpose and not specifically created for viewing in a limited reolution medium Those videos which were created for the limited resources of the net were were often created by website designers with little video experience. (see sidebar "Ten Tips to Better Web Video")

With GUI-equipped internet being only a few years old, website designers witho only a few years experience in both business and technology cannot be expected to offer proficient video production qualities. Yet it is broadcast televsion with which this media is competing and viewers will not long forgive production techniques, like those found in Novel's XXX small business site, where a talking head spends a whole XX seconds to introduce a half dozen more talking heads who will tell us why we have to buy Novel products.

Streaming video as a business requires a steap learning curve both in terms of technology and content and any player who wishes to compete in this high potential market is going to be running a long, uphill race.

Add to that race the confusion imposed by incompatible products. A VDO movie file will not play with a RealVideo browser plug-in. Each brand requires its own player to be downloaded by the end user, taking about 15 minutes out of the end user's curiosity path. In order to eliminate this tragic waste of time, the browser companies and others are attempting to forge server software which will allow all brands and even some additional forms of media to play without the need for brand-specific downloads.

WebCRs

These server products - in development and beta - like Microsoft's NetShow, Netscape's XX, Oracle's Video Server, SGI's MediaBase, Digital Equipment Corporation's XX and XX from IBM, are aimed at the broader purpose of easily incorporating video as one of many media elements. Like VCRs, with various end user features, these "WebCRs" offer the promise of a standardized environment upon which all developers can build. But just the sheer number of players in the server field lead one to believe that here, we will find yet another level of incopatibility, adding to the ever increasing complexity (see sidebar: Multimedia Server Roundup)

The bottom line in this entire enterprise is bit rate and you can't for the forseeable future, get enough of it to make real, 19" sized, broadcast quality television practical on a web. To achieve a threshold level of acceptable television quality on a web, resulting in a sharp, yet pixilated, full screen image with 24 bits color and 30 frames per second requires at least 512 ****

Because this quality of video requires XXX bits per frame, or a throughput of XX mbps per second, only enterprise intranets, equipped with XX connections will support it. While many large corporations have or plan to install such levels of interplatform commujnications, not many of them require high resolution video at each terminal, making this tier a high price play for any developer.

The requirements of supplying thousands of simultaneous video streams with multiple servers increase in cost exponentially with the resolution required. Assuming the target capacity of a large corporation website with 2,500 simultaneous viwers, streaming on 28.8 modems (or an actual stream of 20 Kbps, would yeild a peak bandwidth of about 50 Mb per second. A T-1 line intranet such as that used by GE Capital, with an intranet population of XX, can only carry 1.5 Mbps. To serve even the most minimual emulation of a small cable station, therefore, would require several proxy servers. Obviously the current state of the art forces us to compromise for the time being.

One obvious compromise is to limit the scope of the broadcast to a smaller network, such as that required by a large corporation to distribute training and management communications media. Such intranet applications offer the fastest opportunity for proving the profitability of high quality streaming video. Traditionally, however, corporate video has been motivated by top-down technology flow where the corporate producer emulates the broadcast producer, not the other way around. Marketing intranet video to corporations will be a challenge.

At present, there seems to be a two-tiered market developing around the aspect of dedicated server software - some products offer two products in their bundle, the compression software and the player software. Other products offer a third part, the server software.

For broadcast netvideo, where many people are watching the same video at the same time, server software is certainly a requirement. For video-on-demand, where the total user base for a particular video is but a fraction of the broadcast demand, many will say there is no need for dedicated server software. The difference is important, because those companies offering server software, usually do so on a per-user licensed basis with the cost rising exponentially as the user base increases beyond 50 simultaneous streams.

While reviewing these products, you should analyze your needs for simultaneous streams. A good way to do this is to use one of the non-server products to get some experimental video on your site and start seeing if you develope problems. If not, you may save yourself thousands of dollars.

Tiers of Quality

There seem to be two tiers of streaming video products forming; the high quality - high priced server models and the novelty quality - low or no-priced models. More products enter the field daily and now that Netscape and Microsoft have entered the field, the shake out will certainly follow.

The top tier products offer four product elements; a means of acquiring various media and integrating them for multimedia playback, a means of compressing the audio and video into some networked file, a server distribution software and a client player. No one product offers all of these services (EXCEPT VDO?).

Microsoft's NetShow (beta 2 currently downloading), offers a wide assortment of media integration tools to create audio-supported slide shows, for instance, but does not offer a server which can dynamically adjust the quality of the playback to match the bandwidth of the client.

VDO offers the most aggressive server, with a high degree of dynamic bandwidth matching, but does not offer a means of integrating still frames with timed audio triggers (also called URL-flipping).

Software that cannot conform to meet bandwidth necessitates replicating the media file to match whatever bandwidths are anticipated. If the end user is limited to a corporate network with defined client capabilities, this server software may be a good price choice, but if you're serving the entire Internet, your choice may require a wider expectation of client capabilities.

Most manufacturers in this tier are offering limited versions of their servers, usually with only one or two stream capability, for free or very little cost. Prospective users, wishing to test the capabilities of a system prior to full implementation, may therefore try before they buy.

Quality, therefore, is still defined by a broad spectrum of criteria that do not match from product to product. The goal of all high quality providers, however, is to match the expectations of an audience which demands high video quality and innovative interaction that goes beyond broadcast TV. The quality push is quickly separating the players as many lower cost products fall to the sediment of novelty products while the serious players attempt to distinguish themselves with products that justify their higher cost by providing streams that clients will pay for.

The lower tier products, offered free or at little cost, offer real advantages for the small web developer who wishes to enhance his current sites with video and prepare her customers for advanced offerings. Just like the "free-taste" software vendors, the web developers may offer their customers low initial entry and see if the user interest supports further investment.

These products, like VivoActive and RealVideo are easily obtained, quickly integrated and practical for short term application. They do not, however, offer sophisticated features like hit logging or tracing, but these features can be added with other products if needed before the developer requires upgrading to the high quality streaming software.

The Compressors

All of the software in this test group assume that the developer will be bringing his media to the program from a pre-edited state with the most common file types being WAV and AVI. Curiously, Adobe Premiere, a consumer grade nonlinear editing package, is assumed to be the editing system of choice, though most serious video producers work in Avid, D-Vision or Video ActionNT to create high quality video. This assumption stands out in the video production domain as an indication that the software manufacturers have not adequately researched their customer base, or that they are modeling their business plans for a consumer grade audience.

The quality criterium for compression software, capable of developing A/V files what will look good yet not tax the bandwidth, is the amount of tools offered the developer for online adjustment of the resultant quality. This is best offered by giving the user a side-by-side pair of windows with the source video and compressed video next to each other. Such a design is offered by VivoActive and RealVideo. Other plusses are controls for adjusting audio and video bandwidth, color density, audio fidelity and frame rates with some method of computer-assisted analysis of the results and reportage of suiggestions for improving those results.

We found that VivoActive offered the best controls over compression with a useful set of help windows that popped up to warn the compressionist of possible problems at the client side. RealVideo's compression software came in a close second in this regard.

If a product does not offer a server with dynamic bandwidth capability, the developer will have to create multiple versions of the video file to suit each end-user's bandwidth. A good compression program is necessary, therefore, on such packages, so that the file is conformed with as high a quality as possible for the given band.

The Test Platforms

We obtained test footage for this article using a clone, dual Pentium 200 machine equipped with a Digital Processing Labs (DPS) Perception video acquisition board, capturing a Sony PVW2800 BetaSP source in three-cable, component mode at 6 MB/sec.with audio captured in mono at 44 khz 16 bit. These clips were stored on three Seagate 9 gigabyte Cheeta drives. Using VideoActionNT nonlinear editing software, a test clip was converted from the DPS proprietary PVD format to AVI format as an uncompressed 370 x 240 AVI file.

This file was transferred on an ethernet 10-BaseT to two test server computers - an Intergraph TD-400 (dual Pentium Pro) equipped with an Intense 3D video card and a Tri-Star (single Pentium Pro) equipped with a Trident video card. Both computers were installed with Windows95, WindowsNT 3.51, Windows NT 4.0, and Windows NT Server. For web transmission, the servers were tested on an in-house 10-BaseT network, then on an ISDN loop through the local telephone system and back to the lab, then using a standard 33.6 modem connection in-house. As a cross reference, each product's sample videos were also examined at their respective websites.

Client testing was done on two computers, running Windows95; a clone 486 with a Trident video card and a Canon Innova Pentium 166. Similar modems were used for each client computer.

As might be expected, the quality of video transmitted over faster bandwidths was of higher quality. Products that used UDP protocol provided a more consistent stream of media than products relying on TCP/IP protocol.

Because TCP/IP is a highly reliable protocol, designed for alphanumeric data transmission, it is designed to stop and correct itself whenever data is lost When TCP/IP stops to correct lost media data, the stream of data is halted, causing annoying interruptions in the playback. In media transmission, lost data is not critical, so the halting of TCP/IP to achieve perfect data transmission is not desirable. UDP, however, is an unreliable protocol which sacrifices lost data in favor of maintaining a continuous flow of whatever data it has. Products like Microsoft's NetShow On Demand Server, and VDO's Server, utilizing UDP, and capable of dynamically adjusting their data flow to the client's minute-to-minute reception bandwidth, provided better quality in real world tests because bandwidth is never a fixed figure and rarely achieves the maximum rate as suggested by the modem category number.

Dynamic serving aside, we found that there was very little quality difference between the method or speed of transmission, from 33.6 to full 10-BaseT, between one product and another. If product A had a good looking video at 28.8 and a twice as good video at ISDN, then product B tended to have about twice as good video between its 28.8 and ISDN streams. With the exception of dynamic streaming, therefore, a product's quality tended to improve at proportional increments with all other products as the bandwidth was increased. The key, therefore, is to create as high a quality file as possible on the server and to match the server to the client as closely as possible.

Nonlinear Editors

Many of the products in this test make passing reference to or (in the case of Motorola's TrueStream) rely on Adobe Premiere as some sort of industry-accepted nonlinear editing program. Most video professionals have abandoned Premiere in favor of more robust nonlinear editors like Avid MCExpress for Windows, D-Vision, In-Synch's Speed Razor and Star Media's VideoAction NT. When we noted this fact to representatives of the various manufacturers of streaming software, fast-note taking took place, indicating a willingness to learn, but a heretofore novice understanding about the video production side of this market.

Streaming video, if not now, then soon, will be a highly competitive, quality-aggressive industry. Those succeeding in it will do so because they are the best video producers, the best interactive designers and the best website developers on the planet. With due respect to Adobe and its fine consumer level editor, it is not the first choice of professionals. To all the manufacturers' credit, efforts are being made to address the need to make their tools compliant with the higher end video production tools.

VXTremely Interesting

VXTreme's products are Webtheater 2 Producer, LiveStation, Server and Client.

Webtheater 2 Server

Advanced users will be pleased to find that the server allows for URL page flips and Java aplets to be incorporated in media files, which allows for audio and text messages to be displayed during playback.

Webtheater 2 Client

Vxtreme's client software is one of the most robust of the packages reviewed here. A right-click of the mouse will offer a configuration option for selection of UDP, TCP and single or multiport HTTP. When it starts, the client can be set (CHECK) to request the user to select 28.8, 56 or LAN. It's screen emulates a VCR with controls for Play, Stop, Pause, FF and Rewind. When each command is clicked, a superimposed feedback message, such as PLAY, appears on the lower portion of the screen, further emulating the VCR look and feel.

Client will play on a Windows95 or NT 4.0 machine (no NT 3.51) equipped with at least 16 MB RAM and 6 MB hard drive space. The user must be equipped with Netscape 2.0+ or Microsoft Internet Explorer 3.0+. Mac and Solaris versions are in preview, but will require Netscape 3.0+.

We found the audio quality of the client to be superior to the video quality. Almost no pausing was heard, even when a message popped on the screen, saying, "Network Congested." The video quality seemed to take second place, with significant pausing that made a 28.8 transmission during a busy day look like a slide show. Rather than opt for pixelation, Webtheater 2's compression tends to blur the video.

More information can be found at http://www.netpower.com/webtheater/webtheater.html. Samples of both VXTreme and Vivo files may be seen at http://www.netcom.com in both 28.8 and 56 bitstreams.

RealVideo Gets Real

Procressive Networks, creator of Real Audio, which launched the first level of A/V on the net, is the player with the most extensive track record and its products look it. Progressive boils its compression module and server software into one bundle, selling it in both intranet and Internet
models.

The Internet model is offered for 10 and 20 streams for $295 and $495 respectively and do not support live multicasting. The 60 stream version does and costs $995. The intranet version comes free for 10 streaams and in 100 and Unlimited versions(PRICE?). All versions come with 1 year of free technical support. Call 800-444-8011 (M-F, 8-5 PT) for information.
Boeing, Cisco Systems and Digital Equipment Corporation are listed as examples of large corporations that are incorporating Real Video into their sites. Links can be found at the Progressive web page under "Intranets."

The other half of the package is the Real Player which now incorporates Real Audio. While a free version can be downloaded at http://www.realaudio.com/products/player/index.html. Progressive is the only vendor who offers a version (Real Player Plus for $29.99) of its player for sale. The Plus version includes a CD-ROM with freebies and 90 days of technical support.

Both versions of the player will work on Windows95/NT 3.X and 4.0 as well as MAC PowerPC, MAC 68040 with FPU and Unix.

Encoder and Server

The Real Video encoder was the first we found with a side-by-side comparison window and is second in features only to VDO's latest encoder. The left side window holds the original AVI file and plays it simultaneously with the right window that holds the encoded version during the encoding process. Developers can, therefore, tweek the results with trial and error precision.

At the beginning of the encoding process, the user can enter title, author, copyright information. Although the encoder will accept any AVI source, it seems to prefer 24 bit RGB color as input rather than 256 color and AVI files should be uncompressed.

The encoding process can be very simple by following the default screen's recommendations, but an "advanced" option allows the user to set a "target" bitrate for the media file by setting the video bitrate, frame rate, audio bitrate and latency (CHECK). A wide range of parameters is given.

When the encoding is complete, a pop-up report is produced showing the results in terms of "target" bitrate and "actual" bitrate. If the results were judged by the software as being less than optimum, a notation is generated, such as, "Try re-encoding clip at a lower frame rate or with "Optimize Frame Rate" option enabled in the advanced settings dialog box."

We obtained such a note and tried the optimizer. It resulted in an optimum frame rate of .86 frames per second! Repetitive experimentation with the encoder kept producing the same message, so we created a clip at 15 fps and ignored the notes. Once the file is encoded, the user simply places it on the server system, writes a link to the file and it's ready to run.

The server software offers advanced routines for managing the site. User logging reports both the protocol used and the duration of the stream for each contact. The intranet version features domain and address access restriction and the System Manager software allows remote configuration and bandwidth control with user-friendly GUIs.

Real Player

This is the most robust of players on the market, with more features than anyone could ever use. Modeled after a car radio, the player offers programmable buttons which can be named after favorite "channels." A scan button, which is programmable for various categories of interest, allows the viewer to scan for live or pre-recorded sources. An arrangement with Timecast, allows users of Real Player to configure a mailbot that will cause messages to be sent, advising the user of upcoming and recently revised media sources in any of several selected categories.

Because developers can configure their media files with varying degrees of audio or video enhancement, the resulting playback can be slanted toward either. This makes Real Video a choice amongst record companies who want the audio to get play priority. You'll note while plahying music videos, that the audio stream is the last to burp, while the pictures may slow down to a crawl. The same goes for live broadcasts. However, tune into MGM's website and view the movie previews. Here the priority is picture and the audio fidelity is sacrificed.

In either case, the quality of streaming on Real Player, running from most of the sources we checked and our own tests, yielded some of the best results, with little blurriness or pixelation and audio that was, at times, AM quality or better.

This is the player you've always dreamed of and worth the price of the plus version.
VivoActive for the Hyperactive Webster

Vivo offers its video streaming software in two parts, Vivo Producer and Vivo Player. It is simple, free and quite outstanding. While there is no dedicated server software, Vivo is limited only in the capacity of the server's ethernet card (for intranets) and modem connections. Designed by pioneer Ollie Jones, who remains the company's Chief Technical Officer, Vivo is everyone's free ride on the video net. Download the software, at http://vivo.com, set it up on your site and get going.

Vivo Producer

Our favorite encoder was the VivoActive Producer. Even beyond the high standards set by Progressive, the Vivo left nothing undone and always exceeded our expectations. Like Real Video, Vivo offers side-by-side encoding with more pop-up help windows. The first window, Movie Info, reviews the parameters of your incoming AVI file, which can be in any configuration, but best at full-screen uncompressed, in terms of color depth, length, screen size, data size, etc.

Starting the encoder program, the user is tempted to press the "Advanced" option. This displays a multitude of options, such as Connection Type, Max Frame Rate, Audio, Higher Quality Frames, Delay and ASF. The Connection Type can be 14.4, 28.8, 56, and ISDN. Because there is no server in Vivo, the user is recommended to make a separate file for each anticipated download and even offers a button that automates the process of creating "all types" at one time, and batching them to a common subdirectory!

Better than a list of limited choices, Vivo allows hte max frame rate to be set as a slider with individual frame increments. Audio may be set at "voice" or FM quality with automatic gain adjust as an option. Use caution with this one since intentional low level segments may get boosted unintentionally.

Vivo even offers an option to render the first and/or last frames of the clip in higher quality. Ollie Jones thought this might be helpful because, "You often use the first frame to advertise the clip on your website and you want that to look as good as possible. You might also want the last frame to look good, too, because when the video is over, it freezes on the last frame while the user browses elsewhere." Nice thinking Ollie!

The Delay function is unique to Vivo and addresses the developer's priority choices in download delay time. Set the Start window to 100% and the Vivo player will wait until the entire clip is buffered in the client's RAM before the clip plays. Set it at 50% to begin playing the clip when half of the data is buffered. Another window lets the developer set the amount of time Vivo will guarantee the clip to play without a pause. Set it to the 30 seconds and the clip will not play unless 30 seconds of it will run unpaused.

Because Vivo depends on TCP/IP protocol which is prone to pausing for correction, Vivo built in this powerful control feature to assure the developer of minimal standards of play without pausing. A superbly designed solution.

Player

The VivoActive Player works as a plug-in for Netscape Navigator 2.01 or later and the Active X control for Microsoft Internet Explorer 3.0 or later. Once installed on a web page, it will work under either browser without additional software being downloaded and under Internet Explorer it will install itself without user interaction. The player can be configured on the web page so that it begins playing as soon as the page is opened or it can be set up to require the user to press a play button.

The VivoActive window features only a "Play" and "Pause" control on screen. When the video is playing, the Play button icon, a right-pointing arrow, is green. When the video is downloading additional data, the button ponts down and flashes yellow and red. No key is shown, informing the user what these changes in color mean, but user will probably get the point after a few pauses.

The quality of VivoActive movies transmits rather well. Using the H.263 video compression algorithm, the scenes tend to gather clarity the longer they appear, so you can achieve better results if you limit camera movements, subject detail and quick cutting.

The documentarion for VivoActive is superb, a fully-illustrated booklet of over 100 pages including a chapter on how to make better web video. Using the instructions included in the booklet, even a novice webster can get her videos up and running.

Motorola's Videostream

Motorolla, a company well known for far reching savvy on all fronts of digital communications, entered the market a bit late with its Videostream line (see www.mot.com/video). Videostream comes in three parts; Producer, Server and Player.

Motorola's Videostream consists of three products: Producer, Server and Player. Videostream Producer currently exists only as a plug-in for Adobe Premiere. Although Motorola intends to soon offer Videostream Producer as an Avid plug-in for use on MACs, and perhaps a standalone version some time in the future, your choice is limited to Windows-based Premiere at present.

Compatible with Windows 95, NT 3X, 4.0 and Server, the Producer software acts as the compressor for converting AVI files into Motorola's proprietary MOT format. AVI files must be created in 176 x 144 format or larger.

After completing an edit in Premiere, the editor must first make an AVI movie, store the movie and close the timeline. Then, the editor opens the movie in Premiere as an AVI file, without putting it on the timeline. By merely clicking File/Export and selecting the MOT Movie option, a pop-up window appears, allowing the user to specify one of two resolutions; QCIF (Quarter-screen Common Interchange Format; 174 x 144) and CIF (Common Interchange Format; 352 x 288) in varying video bitrates (17 - 18 Kbps) and audio bitrates (5.3 - 8 Kbps) (CHECK RANGES). 4CIF, or full screen video and eighth screen format are not yet supported, but under consideration. Help file documentation recommends choosing the 17 kbps default for video and 5.3 kbps for audio, which assumes the end user will be streaming at 28.8.

Once the parameters are selected, the rendering process is begun, running at about 5 frames per second. This process yeilds a compressed file in prorpietary .MOT format, which can only be played on a TrueStream Server. Using a 14.6 MB AVI file, we created a 43 KB MOT file in about 2 seconds.

Of all the compressor modules viewed, Motorola's offered the fewest tools for fine tuning the resulting media file. Dual screen comparison of the original file and the compressed file was not supported and the choices of bitstreams were not as robust as other products in the market.

We discovered that Truestream Prodcuer and Intergraph's VizEffects software were not compatible on the same machine and caused Dr. Watson crashes in NT, suggesting a driver conflict.

Server

The developer copies this file to a MOVIE subdirectory on the server's hard drive and then creates an MTV file, which acts as an HTML pointer for the MOT file. The MTV file is a simple ASCII text file including the server's IP address, port and MOT file name. The MTV filename is imbedded in the HTML text for incorporation into the website.

Motorola markets its server for a fee that is based on the number of simultaneous streams supported. The free demo software supports two streams. Increments of pricing are based on 25 at $PRICE; 50 at $PRICE and Unlimited, at $6,000. Prices are the same for intranet as Internet use. The licensed versions offer monitoring functions and usage statistics generation.

While the server will work on most any Pentium PC, a T-1 line will be required for every 60 streams of 28.8 connections, assuming nothing else is running on the server. The server replicates itself on demand and Motorola informs us it will work equally well with a Linux (minimal configuration: 133 MHz Pentium, 16 MB RAM, Linuz Kernel 1.2x or 2.0x) or a PC running NT Server (minimal configuration: 166 MHz Pentium, 32 MB RAM, NT 3.51 or 4.0).

The demo software for TrueStream comes with several sample MOT and MTV files for experimentation. We tested these on both our test servers and they played back both sound and video, but the video was at double speed. (CHECK) When we attempted to play back our own test video file, we were able to produce sound, but the video would not advance past the first frame.

Contacting technical support, we were suggested to try a number of things including changing machines, re-encoding from the native broadcast video file and changing all the software to recently-posted updates. None of these attempts succeeded and both we and the technical support person were able to proceed further.

Using the supplied MOT files, we operated the server over our test networks and experienced mixed results. Although the first 30 seconds of files seemed to play faultlessly, some pausing resulted at both 28.8 and higher rates. We think this is because the software was pre-loading a certain amount of the file prior to actually allowing the player to stream. Once the real streaming started, the player exhibited video halts and audio echoing that ranged from hardly noticable to totally unacceptable. At one time, Netscape's own warning window came up, offering the user the choice of aborting the stream or accepting halts. Since the maximum quality of the video is determined by the compression entered at the Producer stage, no appreciable image quality was noticed between 28.8 and higher bandwidths. Pixelation is the primary quality criterion once pausing is surpassed and Motorola's pixelation is noticable.

Technical support for Videostreaming is by voicemail message and call back. We had the impression that there were not too many people yet online in tech support, although the person we spoke to seemed well prepared (he may have been one of the programmers) and interested in our comments. When we suggested that there were other nonlinear editing programs for which Producer could be formatted as a plug-in, the tech person asked for some names and we could hear him typing them into a computer!

You can view Truestream files at http://www.mot.com/MIMS/ISG/products/video/server/index.html
which offers a choice of 28.8 or 56 modem playback.

VDO

VDOnet's product group VDOLive is composed of three products - Tools, Server and Client - and a fourth product, VDOLive Broadcast Server, which will be soon integrated into the standard VDOLive Server. Live, currently exists to support live, video broadcasting.

While the client is downloadable for free, VDOnet charges for licensing its Producer and Server software based on the amount of streams that the server will handle (in increments of 5 up to 100 with special pricing above 100). (CHECK)

VDOLive Server, targeted for the high quality web developer, and incorporating a proprietary algorithm, is designed to automatically scale the quality of the media to the client's effective bandwidth. The order of quality sacrifice that is made by the server as bandwidth decreases, is frame size (going from 320 x 240 to 160 x 120), frame rate (fully dynamic from 30 fps to key frames then to slideshow rate) and finally resolution. Audio also scales in three increments - 22KHz, 11KHz and 8KHz - after video quality has been downscaled to the minimum, thereby providing the highest level of audio resolution, a plus for music video websites.

We found, in general, that VDOLive's stream offered the highest level of video quality of any streaming software we tested, although the compromises for 28.8 and 56 tended to drop frames to the point of making this video producer wish that he could see more pixels before less frames.

Where VDOLive's products thrive is in the realm of 128Kbps and beyond and the company's philosophy is to develop a strong beach head in emerging high bandwidth market. Here, VDOnet's developers believe that cable modems, ADSL and HDSL (emerging twisted pair technology that offers up to 10Mbs) will soon make obsolete the players that are limted to the 28.8 "standard" common on today's sites.

VDOnet was originally developed in Israel, and is now aggressively supported out of VDOnet's Boston and Palo Alto headquarters. Demo versions of the older 2.1 software can be downloaded for free, with two streams supported at 128Kbps bandwidth with a maximum of 60 seconds of total playback.

VDOLive Tools

Because the server dynamically scales the media quality to the client's available bandwidth without developer intervention, the need for elaborate tools in the encoding software is reduced.
VDOLive's encoder is sparcely featured, with much of the functionality automated below the level of developer concern. Only one media file need be created to serve the needs of all end users, regardless of client bandwidth.

Working as a plug-in for Premiere, as a stand-alone module or as a Windows or PowerMac application, Tools accepts AVI files as raw material, preferring uncompressed, 320 x 240, 24 bit color files with 16 bit 22KHz sound. MPEG1 and Quicktime files must be converted to AVI prior to use in Tools. Tools also features a capture module which will interface with any Windows or Mac compliant media capture card.

Once the AVI file is brought into Tools, the user can accept the default encoding screen and proceed to convert files either singly or in batched projects. Projects can be saved under their own names prior to processing and recalled at quitting time, allowing the computer to batch process overnight. Because Tools makes a file that dynamically sizes itself, the process of encoding is longer than encoders that create specific bandwidth files, like VivoActive, taking about 5 times run time to finish, but the task need only be done once for each file.

An advanced button in Tools, brings up further tools: The Maximum Bandwidth slider allows the user to select a setting from 14 to 512Kbps. Mode allows any of four options for video frame rate (VDOLive movie = 30 fps, Flip Book = key frames only, Story Book = synchronized frames and Audio Only). Copyright allows the developer to add a line of text that appears in the right clicked "details" choice of the client player.

After these selections are made the developer initiates the encoding process and is then presented with a unique window that allows for play back of the encoded file with various bitrates emulated. This is a very valuable tool for examining a files playback properties in various anticipated environments. If the developer is not satisfied with the results in any particular bandwidth, she can return to the encoder parameters or even to the nonlinear editor to adjust the AVI file and re-encode before committing the file to a site.

Server

VDOLive Server can run under NT and Unix and requires at least a Pentium 166 with 32MB RAM. The Control Panel allows you to limit the maximum number of users and the maximum bandwidth per user. By adjusting these two parameters, the developer can tune his system to the needs of his audience and the capabilities of his bandwidth. For instance, one developer, limited to a single T-1 line, may wish to limit bandwidth to 28.8 per user, but allow the maximum licensed amount of streams. Another developer might use the T-1 to limit the users to few, while maximizing the bandwidth of each stream.

The server runs as a daemon in Unix and as a service in NT, allowing for automatic start whenever the server is rebooted. This reduces the amount of time required to get back on line after servicing the host machine.

Server also supports logging of users and usage information, including user email addresses if the user has not specified "annonymous" in the player's setup. This is usefull in building electronic mailing lists.

Player

When initializing the player, the user is given a registration screen, requiring name, company and email address, all of which may be made confidential by clicking the "Annonymous" check-box. If the checkbox is not selected, the user's ID is transmitted to the server.

The player installs as a stand-alone application, Active X component and as a plug-in for Netscape. The player, whose standard operating mode is UDP can automatically switch to HTTP to serve a user who resides behind a firewall.

The user may resize the window by double-clicking on the screen and if the machine supports DirectDraw, the file will play full screen. DirectDraw may improve playback if the end user's video card is DirectDraw enabled. Such cards as Matrox's Rainbow Runner will transfer video processing from the CPU to the video card, thereby reducing CPU decoding overhead and improving video quality.

The player also features a red, warning screen that appears whenever problems exist with the data stream or bandwidth. The current player has only Start and Stop controls, but further revisions will offer full VCR controls, which VDOnet developers feel is the developing standard for streaming players.

As video producers, we appreciate the efforts of companies like VDOnet, which allow us to encode our files today for maximum anticipated bandwidth, while allowing us to serve the low-end clients that generate the majority of cash flow today. This reminds us of the days when we were stuck with 3/4" videotape masters and the world went to Betacam.

Vosaic

Sidebar #1

The Late Arriving Giants

Any serious consideration of what streaming software to buy must take into account the plans of the giants in the field, especially the two who dominate the market share of browswer software. Just as there are over 2,000 browsers in commercial distribution, only two - Microsoft Explorer and Netscape Navigator - have any market share worth considering.

What would be the commercial effect on a player like Progressive Networks (maker of RealVideo) if Navigator's next rev had a proprietary Netscape video player built in? What advantages will VDOnet have as a result of being partly owned by Microsoft?

Past incorporations of new media types by both Netscape and Microsoft have tended to be third party compliant, if not third party acquired. Added to this analysis is the rumor that Apple is working on a streaming format for MAC users. Knowing that MAC users would rather run through hell in gasoline-soaked jumpsuits than be brand-disloyal, what lies in the eye of Apple's engineers?

Microsoft's coming NetShow, Netscape's coming LiveMedia product, and Apple's coming Streaming QuickTime, when and if realized, offer some startling suprises, both good and bad. At present, we are not impressed with NetShow's demand for high technical acuity and lack of user friendliness. Those developers, for instance, who designed for NetShow Beta 1, have now learned that they will have to rewrite their HTML pages to conform to Microsoft's change from ASR to ASX redirectors.

Microsoft's entry into the streaming market is highly aggressive, however, with ambitions to incorporate virtually every codec standard, perhaps quickly obsolescing some of the work that smaller companies are investing in the low end of the streaming market. This scenario has appeared before when large software companies offer inclusives that smaller companies regard as their revenue stream.
Using a common Active Stream Format (ASF) to archive multimedia content before streaming, allows NetShow the ability to synchronize still images with audio to create interactive learning experiences (see Street Inc.'s product for web training). As Microsoft gains credibility, the smaller players, like VivoActive, which has added a "make ASF file" button to its Producer control screen, are adapting to anticipated user demand.

In addition, Microsoft is incorporating compatibility of most of the major streaming software companies into its framework. At present, such compatibility exists for VivoActive, VDO and (ADD).

Microsoft considers its only competition in this game to be Progressive Network's RealVideo, which Microsoft acknowledges as the pioneer of the market, and so far, nothing has been publically announced regarding a compatibility between RealVideo and NetShow.

A loss-leader for Microsoft, intended to increase usage of NT Server, NetShow's acceptance will be enhanced by its complete compatibility with Windows NT Server, for which it is fully optimized. This will allow end users to scale their operations flexibly from a single Pentium processor (which Microsoft claims can handle up to 1,000 streams of data at 28.8 Kbps) up to multi-processor DEC Alpha machines with scores of Seagate 23 gigabyte drives.

Like Godzilla, walking through Mahnattan, even if Microsoft tiptoes through the landscape of video streamers, someone is going to get flattened and it is our fear that the small players (Vosaic, VXtreme and maybe even Motorola if it doesn't get into the market faster and deadlier than it is) will be scrap bricks by the end of the year. Small players need to be there and survive a little longer for the market itself to mature and become marketable to the web-unsavvy.
Given time, they will most likely retreat to profitable niches, or get acquired. A bloodbath at this time is not good for anyone, including Microsoft.

MORE NEEDED

Sidebar #2

Ten Tips to Better Net Video

As we tested and reviewed the leading products in video streaming for this article, we searched the web and found a lot of badly produced video out there! Video that's killing sales and PR. Dreadful stuff. Here are ten tips from pros with over 20 years of broadcast and corporate video experience to improve your bottom line results in using video on the net.

1. Don't use videos that were produced for TV monitors. Video for the net has to be produced from scratch or at least re-edited specifically for the net. Some companies try to save money by taking existing video programs that were designed for 19" TV monitors and converting them to net format. For many reasons, this does not work. Net video reduces resolution tremendously and what looked good on a TV looks embarassingly bad on the net. Trying to save a money by rerunning corporate videos in a net environment is throwing money away. Net videos must be designed for net limitations. Here are nine more tips related to making your net video look good.

2. Make the Titles Big. In television, this is called a bottom third. You probably can't read it. On the net, let's call this a bottom half and make the letters big, big big.

3. Avoid boring talking head shots. Often the lips won't synchronize anyway and unless the face is famous, who wants to watch it? Also avoid big close ups, so the bottom half won't cover the person's face.

4. Avoid shots with small details. You cant show the parts of this speadsheet on a net because the pixel resolution is not fine enough. Use a close up instead.

5. Make the Show Short: Already I've found a plethora of boring lecture videos that run as much as ten time longer than their content deserves. Hey, this is the MTV generation you're playing to. Give them hot editing. Give them music that stings with the video like this!

6. Use a digital camera and edit digitally: Forget expensive broadcast quality systems. Analog video is like scratching your right ear with your left hand. Use a good digital prosumer camera like the Sony DVX1000, connect it with a firewire to your computer using the DPS Spark card, edit nonlinear and conform directly to a streaming video format idealized to 28.8 modems. If you don't understand this tip, call a video professional with web experience.

7. Open and close your clips with high-quality still frames (VivoActive Producer allows you to make these automatically). Although your run-time video may look low grade, the user will be tempted to start something whose first frame looks great and the last freeze frame will leave a good parting impression.

8. Use lots of quick cuts to illustrate your narration. If the show bogs down, it will cycle the quick cuts like a series of stills, thereby looking more intentional than if you'd used long dissolves or wipes that may, themselves become ugly freeze frames when the stream pauses