Classroom Connect '98 Conference:

Introductionary Videoconferencing Concepts

by John Kuzmich, Jr.

November 11, 12, and 13, 1998

Stand-alone high-end videoconferencing systems for meetings, with large screen (31" and larger) and a price of $40,000 and up are fairy common in major corporations. Stand-alone consumner-entry level videoconferencing kits now sell for $199 and up. Their prices keep dropping, their quality keeps improving and most significant, standards have evolved over the past year so that similar products from different vendors can communicate with each other. All include a camera, an ISA or PCI video capture board to install in your PC, and the requisite software. Some kits have a built-in modem or ISDN adapter. All provide a whiteboard utility which is a shared window in which both parties can make notes or paste graphics. The best products can produce adequate video at 352 by 288 pixels, a format know as CIF (Common Interchange Format) running about 15 frames per second. None could replace most face-to-face meetings. Setup is another area of concern because it is not a simple procedure and many times it takes considerable effort to get the products running at full potential. If you are looking for television quality, 30 frames per second, you will not find it without considerable expense.

Four Access Paths To Videoconferencing:

The more bandwidth you have to send the video and audio from one desktop to another, the better the image quality. Phone lines provide the least bandwidth whereas LAN connections offer the most.

1. Directly over the phone line with H.324 standards compatibility. This is a direct connection between two modems. Because you don't share the line with everyone else on the Net, video quality is often significantly better. Frequently, you can obtain video considerably better, but still a bit jerky at about 8 fps. H.324 governs modem-to-modem video. Note: you must pay for long-distance charges.

2. Directly oven an ISDN connection. A direct link between two ISDN adapters which offers almost four times the bandwidth of a modem-to-modem connection and improves video quality substantially. You can run up to 10 fps with a CIF image and 15 fps with a QCIF image. ISDN is a relatively costly option. In California with Pacific Bell, an ISDN connection can be about $35 a month per residential account. Long-distance charges are double per-minute phone charges of analog phone services. There is also an extra fee for a 128-kbps connections because it uses two lines. Getting ISDN installed can also be expensive. An ISDN modem costs about $300.00. The H.320 protocol standards compatibility that ISDN uses is incompatible with H.324, an ISDN system can't connect directly with a PC using a modem. On-screen details such as glasses, necklace, hair style are possible from ISDN.

3. Over a LAN. Though Ethernets can transmit a whopping 10 to 100 MB per second, this rate is also divied up among the dozens or even hundreds of people on the LAN. LAN videoconferencing uses the H.323 standards compatibility protocol, so if your LAN is hooked up to the Internet, you can communicate with telecommuters, clients and suppliers over the NET as along as your LAN allows you to push video through its security firewall. LAN-based video yields detail such as laugh lines around the mouth. But don't expect TV quality at 30 fps.

4. Through the Internet. This is the cheapest way to videoconference. You will need a camera (videocam), an ISA or PCI capture card for PC's, and download some conferencing software such as NetMeeting). The standard for sending and receiving video over the Internet is called H.323 standards compatibility. You can get a 176 by 133 pixel image (Quarter CIF or QCIF) and only several frames per second which is very jerky. The smaller the image, the clearer the image. Internet traffic jams can degrade that frame rate to zero and garble the audio beyond recognition. The connection is based on the slowest modem and frequently the blurriest when connected over the Internet using regular phone lines.

Videoconferencing software products require a lot of drivers among other software components. Bandwidth is a very important ingredient. Pick the best path available to you. Higher-bandwidth connections like DSL, cable modems, and 384-kbps ISDN should make desktop videoconferencing more practical in the next year or two. Of particular note is a new technology called XDSL that utilizes your present analog telephone service with the capabilities of 256 kps and greater, plus allows you to use your telephone with two simultaneous transmissions (fax, voice and/or Internet). The good thing about XDSL is that the long-distance charges are based on your analog phone service and the monthly service has only one charge with unlimited "air" time. In Denver, US West is offering this service at only $25.00 a month beginning in January, 1999, with a free $300 modem, as well. 256 kps is twice the transmission rate of ISDN. Presently, US West XDSL service costs $59 a month.

Benefits of Videconferencing!

1. Save money on transportation and related expenses. Great for guest lecturers and student interaction with other students.

2. Save employees' time by not having to travel between schools for meetings, etc.

3. Offers instant feedback with a personal touch!

4. Leverage existing investment. If you install an H.320-to-H.323 gateway on your LAN, anyone on the network with microphone and a copy of NetMeeting (which is free) can join a group video-conference on the speaker phone.

5. Use Existing Bandwidth Management Tools. Check to see if some of your bridges and routers already came with some traffic-prioritizing tools that could be used to improve video frame rates.

Before Purchasing Videoconferencing: Establish Videoconferencing Goals!

1. What training/support is needed?

2. What hardware is needed?

3. What software is needed?

4. How will student growth be evaluated in the classroom?

5. How can these new videoconferencing experiences be shared with other teachers, administrators and parents?

6. What content areas will videoconferencing affect in the classroom?

7. What resources are available to help you accomplish your goals (i.e., grants, libraries, bookstores, web sites, staff members?

Robert Scoble and Kevin MeMenamy Interviewed: Videoconferencing Experts!

On November 7, 1997, and May, 1998, I interviewed with Kevin McMenamy and Robert Scoble, videoconferencing experts affiliated with Winnov Corporation, a leading video capture card manufacturer. Please note that Robert Scoble is no longer with Winnov.

Are there any differences between video telephony and video conferencing?

None that I know of. If there are, they are pedantic differences that no one really cares about.

My definition of telephony refers to either audio or video conferencing I define it as the science of transmitting voice, data, video or image signals over a distance greater than what you can transmit by shouting!

Most of the market now is concentrating on H.323 (which means IP to IP conferencing) and H.324 (which means modem-to-modem conferencing). Most educators should be interested in H.323 conferencing, since most of their students will be on the Internet. In fact, at the latest Microsoft Internet Explorer Launch Event a teacher (who was in San Francisco) called one of her students at home (I forget where he was, but he could have been anywhere in the world) and helped him with his homework (they did application sharing and he showed her a video of a science project he was working on).

Video Conferencing Standards

Standards rule in the world of telecommunications. The controversy of the 1994-95 era is over. ITU standards have won the skirmish. Two years ago, all the attention was on H.320. Today, we have H.324, H.323, H.322 and H.321/H.310, as well. H.324 covers the POTS network in the home. H.323 covers IP networks in the workplace. Both are keys to a rapid market buildup.

Don't make H.323 just for businesses. More home users are using H.323 today than are using H.324. Why? Because H.323 lets someone call anywhere in the world for free. H.324 makes them call long-distance and hook up to a modem. For instance, if you need technical support from Winnov, you can call me at 208.203.3.120 using NetMeeting (I leave it on 24-hours a day and it'll automatically answer) but you'll never be able to get me with H.324 or H.320.

You can NOT separate out H.323 and H.324 by types of users. I know some corporate users who are using H.324 because they want to make sure their conversation is as secure as possible, and by using modem-to-modem they are less afraid of some hacker listening into their phone call than if they used H.323. Also, H.324 (modem-to-modem) gives somewhat better quality (bigger pictures and faster frame rates) than H.323. That's because H.323 needs to wrap each packet with additional overhead telling it where to go on the Internet. H.324 doesn't need that overhead, because it only needs to find the other modem, not go through five routers on the way to China.

H.323 is more convenient, is cheaper, and is more ubiquitous (every Windows operating system from now on will have H.323, not H.324).

H.323 (Videoconferencing over IP), or H.324 (Videoconferencing over POTS), or even H.320 which is what some products support (Videoconferencing over ISDN) which indicates that the application they are using for v/c is proprietary. This means that you can only communicate with another videoconferencing solution of that brand of software, such as Summersoft and VDOPhone.

That's not true. Summersoft, and VDOPhone do NOT comply with the H.323 spec. These are truly proprietary videophones. If you want to call a Summersoft user, for instance, you need to have a Summersoft videophone. VDOPhone users can't call Summersoft users.

Therefore, a H.323 user can communicate with any other H.323 compliant videoconferencing application, such NetMeeting, CU-SeeMe, and VideoPhone by Connectix.

This is true, sorta. Enhanced CU-SeeMe 3.11 is the first CU-SeeMe version that is H.323 compliant. VideoPhone by Connectix is NOT H.323 compliant (at least not yet). Only Intel's Internet Video Phone, Enhanced CU-SeeMe 3.11 and NetMeeting 2.0 or newer is H.323 compliant today. There are lots of others that will come out over the next six months.

The H.3xx standards are umbrella standards defined by the ITU-T (or CCITT) that describe methods to allow Videoconferencing terminals from different manufacturers to interoperate. H.320 includes H.261 (video compression), and several audio standards (G.711, G.722, G.728). H.323 includes H.263 (a new video compression standard), and several other audio standards including the ones listed above.

You can find some basic videoconferencing information in several places on the web listed below:

http://techweb.cmp.com/nc/704/704buyers.html

http://www.visc.vt.edu/succeed/videoconf.html

If vendors are pretty much following the standards at this point in the game then why is VDOPhone and Summersoft are not following these standards?

Because standards force vendors to stay with "approved technology" and this field is simply changing so fast that standards can't keep up with the technology. For instance, if you want multiparty video, you'll need to go with the proprietary version of CU-SeeMe today. If you want better frame rates or features that aren't on NetMeeting, you might need to go with a proprietary videophone. Videoconferencing is a field that is undergoing MASSIVE change. A year ago all I could get was about two frames per second on a 28.8 modem. Today we're getting more than 15 frames per second. The market is moving toward standards, though, mostly cause Microsoft has endorsed the H.323 standard. If you want to be able to call a NetMeeting user, your phone will need to be H.323 compliant. Personally, NetMeeting is really pushing the market because Microsoft will deliver about 100 million customers next year alone. Third-party vendors can't get even a million people to adopt its technology. So, really, the only standard that counts in the marketplace is the one that Microsoft endorses.

Low End Consumer Analog Video Conference

There is no compression/decompression of video streams. Can I say Internet (IP) Videoconferencing is solution targeted mainly at consumers.

No, IP Videoconferencing is NOT just aimed at consumers. Actually, a huge user of H.323 is corporations. Think about it, everyone inside nearly every company in the world is on a IP-based network. Why would they buy anything BUT H.323? (Hint, they wouldn't, and they aren't).

In reality, low end products can claim 30 fps video display, but if they try to send the video across the Internet, it probably only gets about 4-6 fps.

Be very careful here. Frame rates are over hyped and misleading a LOT of people.

Almost every camera and video capture card company quotes frame rates for capture to disk ONLY!!! Be very, very wary of any company that says they can do 30 frames per second. Almost always it's a lie. There are a lot of variables to Videoconferencing, and only one of the considerations should be frame rates (quality, audio quality, features, filtering, processor usage, should be among the others).

OK, let's take NetMeeting 2.1 RC1 (I've been doing a lot of tests on that).

NetMeeting 2.1 on MMX machines, even over a 28.8 Internet bandwidth can achieve more than 15 frames per second now. Keep in mind that when you are talking about H.323 you need to know about what bandwidth you are getting: one problem, though, is that on the Internet you NEVER KNOW what the bandwidth really is. Here's some things to think about:

Remember, when connected to the Internet there are at least THREE gating points.

1) Your computer's connection to the Internet (or a corporate LAN).

2) The bandwidth on the Internet itself (who knows what that is).

3) Your partner's connection to the Internet (or your LAN).

Even if you have a 100 megabit connection and are on a 100 megabit network, if your partner is hooked up with a 28.8 modem, you will only have 28.8 kilobits per second of bandwidth. (NetMeeting can use up to around 600 KB of bandwidth, so you can see that in such a situation you are limiting the performance of SOMETHING).

Getting everyone onto ISDN bandwidth (128 KB per second) or even better, cable modems, which are now starting to see usage in homes (up to 10 megabits per second of bandwidth) will give much better H.323 video quality.

If a direct connection (point-to-point) is used for video conferencing, is the fps generally closer to 15 fps?

What really needs to be said here is that H.324 has bigger sizes and better frame rates and more consistent bandwidth, but that it isn't enough better than H.323 anymore to really cause anyone to use it. One cool use of H.324 is the new 8 x 8 via TV phone which lets you buy your computer-illiterate mother a videophone that she can hook up to her regular telephone and television and then you can call her "set top box" from your computer with H.324 software (by the way, I recommend Smith Micro's Audio Vision software).

High End Video Conferencing

High end video conferencing provides on-board compression and decompression of video and audio providing transmitted frame rates of 15 fps at CIF (352 x 288), (30 fps at QCIF - 176 x 144) over ISDN (128 kbps). Connectix claims 640 X 480 pixel resolution Color CCD from its Color QuickCam.

That's incorrect and it's why this industry is getting bashed for over hyping its products. The CCD itself is capable of 640 x 480, but I guarantee you the Connectix is incapable of delivering much more than QCIF sized video 176 x 144. If you try to push the Connectix to do bigger frame sizes it starts sucking your processor and then NetMeeting starts behaving irrationally. In fact, on Windows NT the Connectix causes your entire system to cease working when sending large video sizes.

NetMeeting, on a 28.8 modem line on a clean Internet connection, with MMX computers can do the following:

SQCIF (between three and twenty frames per second).

QCIF (between three and fifteen frames per second).

CIF (between one and nine frames per second).

Remember, there is NO SUCH THING AS A FRAME in Videoconferencing!!! To get the compression ratios that are required to send video over a 28.8 modem, they are using interframe compression techniques. So, if you don't move, NetMeeting doesn't need to send new data and the frame rate goes way up. If you move only your finger it will only need to send that data, not the data for the wall behind you, or your face. Frame rates will still be very high. If you move everything in the image, however, NetMeeting will need to send much more data and that is why I give a range for frame rates. (By the way, these rates are apparent (subjective) rates, because in NetMeeting there is no way to objectively measure frame rates.)

Do the different resolution variables correlate to a higher video transmission rate (fps) or does the quality of the picture photo go up as the resolution rate goes up?

In NetMeeting, you can have any two of the following three items:

Fast frame rate.

Large video size.

Sharp quality.

Videoconferencing is a practice in tradeoffs because of the extreme limitations of bandwidth, such as fitting video into a 28.8 modem connection. Does Connectix's Color QuickCam hardware claim mean that it can realististically accomplish this resolution rate with only a point-to-point ISDN connection?

Connectix claim has nothing to do with Videoconferencing.

The main difference here between low end and high end is that it's a codec?

Not really. There are feature differences (see the list at http://www.winnov.com/market_overview.pdf). There are quality differences (the imaging devices in low-end cameras are not as sharp even though on paper they say they are). There are performance differences (parallel-port and USB cameras are much slower and use more processor speed than other cameras).

So high end does better compression/decompression of video and audio streams, which transmits a clearer picture, better colors, less blockiness, and a lower latency than a low-end solutions.

Not really, see above. If you are talking about ultra high-end systems like PictureTel's systems, yes, that's true (they do hardware-based compression/decompression).

Also, low end video conferencing solutions utilize the Internet for the transport of video which is

unreliable, and at one point, the quality could be great, and the next moment, the video could be almost stopped....whereas ISDN is point to point, so you get the same bandwidth - 128 K at all times...

In reality the Internet rarely just stops working. The quality lately has been quite good. The more bandwidth you have the better it is, though.

Differences between high end ISDN vs. lower end consumer POTS solutions....

The quality and the price, as well as, the mode of transport are the main differences. High end ISDN solutions almost always are an add-in codec (card) plus software and camera and microphone. An ISDN (BRI) line has to be installed at the desktop or home, and it will provide up to 128 Kbps output. The ISDN line can also be used for faxing, e-mail, telephone, etc.... The low end POTs solutions use the analog telephone lines in the home and usually consist of a software codec (if any codec), a video capture card and a camera. ISDN solutions can give up to 30 fps over ISDN lines today, whereas POTS solutions generally give less than 15 fps video.

Even ultra high end solutions rarely give more than 15 frames per second.

Hardware Considerations:

Minimum CPU's is a Pentium 100 MHZ speed.

MMX Pentiums do help quite a bit (twice as good a frame rate, for instance, at CIF size with Winnov cards on NetMeeting).

Any recommendations about which 56K modems to use since there is no standard compatibility among 56K modems between 3 COM and Rockwell? 56-kps modems will not replace ISDN as an access method for business applications.

56 K modems don't help videoconferencing since they are 56k downstream only.

Video conferencing capabilities are now routinely bundled with modems and cameras. H.323 compliant audio-only telephony and full videconferencing capabilities are being embedded into the most popular Web browsers.

How can one distinguish Connectix's and Winnov's product features and pricing?

See the URL list. That covers pricing and features. I would also like to define differences in video capture cards (ISA, PCMCIA, and PCI) in terms of their among their hardware capabilities and pricing?

Also covered on the list.

Can you can get between 20 and 25 pfs with most of the PCI-based cards?

Most PCI based cards are slower than our ISA cards because our software is more efficient. That's why Microsoft uses our cards to show off NetMeeting.

When video conferencing point-to-point, you can set frame rates of 15 to 20 fps over a high-quality connection, but more likely, you'll get 10 to 15 fps. Over the Internet, things slow down even more to 10 fps or slower. Are there any articles people can reference about video capture cards?

There are several but none have done a good job comparing them. I don't have any URL's on the top of my head though.

Are there any major differences between the capabilities of video conferencing on the Mac and PC platforms?

Almost all of the videoconferencing innovations have been made on PC's lately because that's where the majority of the market is.

Winnov cards are the only ones that can take advantage of MMX CPU technology?

True

Video file compression is an art. Not all codec's are equal when taken into consideration video capture card software compression capabilities.

True

You can take advantage of any two of the following three setups.

True

1. Fast frame rate

2. Larger video size

3. Sharper image quality

What are your thoughts about video image size in terms of frames per second and image quality?

It all depends on bandwidth. NetMeeting can go up to 600 Kbps. If you had that much bandwidth available, you get the green performance marks, if you had a 28.8 modem, you'd get the red performance marks. This is only on Winnov. Other products would be slower.

Common Intermediate Format: CIF (352 X 288) 5 fps/1fps

Quarter Common Intermediate Format: QCIF (176 X 144) 15 fps/ 5 fps

SQCIF (what resolution is it? I can't remember ) 30 fps/15fps

Connectix claims 640 X 480 pixel resolution Color CCD from its Color QuickCam! It can take a single picture at 640 x 480. It can do full motion video at much smaller sizes. Winnov's new PCI cards can do 640 x 480 at 30 fps.