TechTalks Transcript
Network Digital Video
 Judith Boettcher [JB] |
 Howard Strauss [HS] |
 Joel Mambretti [JM] |
 Bob Taylor [BT] |
February 3, 2000
Audio
• Streaming
MP3
• Download
MP3 (Download
Tips)
JB: Welcome to the CREN TechTalk series for spring of the new millennium, and to this session on Network Digital Video. You are here because it's time to discuss the core technologies for your future campus.
This is Judith Boettcher, your CREN host for today, and I'm pleased to welcome back our regular technology anchor for TechTalk, Howard Strauss of Princeton. As you know, Howard is a well-known, very energetic Web and renaissance-like information technology expert. I may have to change that intro....
HS: Keep saying those things, Judith!
JB: Right! But welcome, Howard.
HS: Thank you, Judith.
I'm Howard Strauss, the technology anchor for the TechTalk series of CREN Webcasts. As technology anchor, I will engage our guest experts in a lively technical dialogue that will answer the questions you'd like answered, and ask those very important follow-up questions. You can ask our guest experts, Joel Mambretti and Bob Taylor, your own questions by sending e-mail to expert@cren.net anytime during this Webcast. If we don't get to your questions during the Webcast, we'll provide answers in the Webcast archive.
A few years ago, my mother stopped by my office, where I showed her the very latest in computer technology. I accessed a Website in New Zealand and in just a minute or two, I was able to display a two-inch-square, 30-second QuickTime movie on my screen. The motion was jerky and the sound was creaky, but I was awed by the fact that I lived at a time when such incredible things were possible.
My mother, who still does not quite understand what I do for a living, could not believe that I was pleased that I could use a computer that cost ten times as much as her TV set to show a short movie of such poor quality. "If my cable TV was that bad," she advised me, "I would refuse to pay them,"
For most of us, things have gotten only a bit better. We understand, of course, that a computer is not a TV set and that the Internet is not a cable TV company. The challenges of carrying high bandwidth video signals across the relatively slow and very crowded networks that most of our computers connect to is a much more difficult problem than TV broadcasting.
Because we understand how hard it is to deliver high quality images and video to our desktops and laptops and palmtops, we've grown to graciously accept what we have. In talking with Bob and Joel before this Webcast, I mentioned that while I was unhappy with network video, I felt the still images on the Web had plenty of resolution. They were aghast! Would I trust my doctor to look at my x-ray films or MRI scans at current Web resolutions? Am I satisfied looking at fine art on the Web, as I am seeing it in a museum? Of course not! We have a long way to go.
But ready or not, the network video floodgates seem about to open. Millions of digital cameras, both still and video, are being sold. A simple video camera for a computer can be had for under $100. Apple offers pretty good video editing software that anyone can afford, and new standards for video such as MPEG4 promise that full-motion, full-screen video will soon be commonplace.
For some video applications, the commodity Internet will not be adequate. For others, the wiring to desktop machines will have to be upgraded. For everyone getting involved in this area -- and all of us will be involved sooner than we think -- there will be dramatic changes. Things that seemed like science fiction -- like operating a complex scientific device remotely -- will become something students expect to routinely do.
The promise of video and high resolution images over the network raises many questions. We need to know what's really happening in this exciting area and how to acquire, edit, encode, distribute and manage digital images and videos. Can we get HDTV at our desks? Is this the end of video stores? And is there a network video application that would impress my mom?
Bob and Joel will tackle those questions and more and give us some hints on how we can all get started in the area of network digital video on today's Webcast of TechTalk. Judith?
JB: Well, thank you very much, Howard. And I will only add that just when desktop publishing came along -- we were all able to publish newsletters -- that we started wondering about the value of common sense. Perhaps as we move into network digital video, we will hear from Bob and Joel about how we can use this stuff as it's perhaps intended to be used in good ways for instruction and research.
Welcome, Bob and Joel, and let me just introduce them to you very briefly. They are both from Northwestern University, Bob Taylor and Joel Mambretti. Bob is the Director of Academic Technologies and Joel is the Director of the International Center for Advanced Internet Research. And while both have a range of responsibilities, they've both been leaders in this area of network digital video. And I'll just add that I've been trying to get these folks to come on and talk to us on TechTalk about this for about a year now, and I think you're finally ready, right, Bob and Joel?
BT: Judith, it's a pleasure to join you and Howard, and frankly, in the last year we've made a lot of progress, so I think this is the right time to do this.
HS: That's what you were waiting for, for this thing to start working.
JM: And let me add my thanks for the opportunity to talk about this very exciting subject.
HS: Okay, maybe we could start by trying to explain what in the world network digital video is and why we ought to be doing this kind of thing. Bob? Joel?
JM: Well, why don't I take the first shot at this and do it in kind of a global perspective, because I think both of your introductions to the subject were quite good in that in part you related this to what general people perceive.
I think that one of the interesting things about discussing digital video on the Internet is it's something that everyone understands immediately. It's not like some of the more arcane topics that you get into when talking about the Internet. But you can tell anybody basically about having high quality video through the Internet that will be available through a PC or a TV or handheld device and people understand basically what that means. They have a common ground for understand that.
HS: Do we understand what high quality video is?
JM: Well, we're going to get into that, I think, in this conversation. I think we'll get much more explicit about specifically what we mean by high quality.
HS: Okay, but what we're talking about today is high quality digital video, then?
JM: That is correct, but we also have to talk a little bit in that context about some of the lower quality video. For example, at some point, we're going to be talking about classes and maybe 30 people in class watching very high quality video. But then there might be three people who are part of that class that simply don't have the access to the capabilities of that, but you want to take care of them also. So they might have to see the video at somewhat lower quality.
JB: So you're saying that basically -- we may be getting ahead of ourselves here -- but people in the same class may well see the same video, but at different quality rates?
JM: Yes, but it would only be, hopefully, a small subset that would see the smaller qualities for specific purposes, but in general, we are focused on very high end, very high quality video. And to some degree, some of this video will be much better than what you see on a TV.
HS: When we talk about this network digital video, do we mean networks inside the university, or do we mean the Internet or what kind of networks are we talking about?
BT: I think it's easiest for most of us who are at universities to distribute and to exploit high quality video when we think primarily -- or at least first -- of our own campuses.
It's actually pretty amazing in that universities are somewhat repeating the early history of the Internet in that right now, universities are in many cases the best positioned sectors of our society to actually be using high quality network video. And the universities may not even realize that yet, but in many ways, I think the universities' campuses during the next two or three years is going to be the test beds. And by that, I mean really a social test bed as to how network video can inform instruction and research.
JM: Well, I would go further than that and I would say they are today the test beds.
BT: I agree with you!
JM: Because they do have, in most cases, a broadband network built up that can actually support high quality streaming video, and Bob will certainly be talking about many specific examples at Northwestern.
But getting back to your question about whether this means campus or a broader area, Bob is right in pointing out it's primarily right now within the campus. However, there are important projects going on which are intended to focus on the wider area, and that's a pretty exciting area, too.
For example, the Internet2 Digital Video Project which envisions information exchange among the major universities, where they can show distance learning classes, special colloquia, or they can show modeling animation of scientific projects. There are many things that can be done as information exchange across the universities -- not only nationally, but internationally.
JB: From what you're saying, Joel, it sounds as if the campuses will be challenged, perhaps, to upgrade their network, their campus networks again. Is that fair to say?
JM: Well, that's actually a very important point and it's particularly important to, I think, administrators of universities. That is, they'll probably be saying to themselves, "Is this yet another technology the IT people are asking us to support?"
And the response to that really is that it's not so much that the IT people are driving a particular technology, but rather what we find is that faculty want to enhance learning and they want to enhance research and they see this as a very powerful tool to enable new kinds of learning and new kinds of research through simulations, through animations, and through bringing certain experiences in the classroom that were not possible before.
HS: Could you talk just a little bit more about some of these applications, some of the kind of things that people would like to be able to do that they really can't do without this kind of thing?
JM: For that, I'm going to turn it over to Bob because he has a number of really good examples.
BT: Let me talk about some of the learning technology applications that have been embraced at Northwestern, and then maybe we'll turn back to Joel for some of the maybe medical research applications that he's familiar with.
We began actually first using what I would call high quality networked video in the spring of 1998. I was able to show some Spanish instructors that spring MPEG1 quality videos streaming over our campus network. Now, MPEG1 quality video is about the quality of video that you would get from a CD-ROM or if you went to an older VHS tape and saw a video. The screen resolution is about 325 by 240 and there's 30 fields per second.
I found that this was the first time in 15 years of showing faculty members on various campuses digital video, that this was the first time when I showed them MPEG1 video, that there's a gut reaction on their part where their eyes lit up and they said, "This is good video! I'm certainly interested in this," other than just kind of impressed that you can do anything at all. Somewhat like your mother, Howard, you know. Impressed that they could do it, but what's the use of it? As soon as I showed them what they could do with MPEG video for Spanish instruction -- and here we were faced with the problem of weekly, 28-minute long episodes of a tele-drama from the Spanish --
HS: This is a Spanish soap opera people were watching?
BT: Yes, it's called Destinos, and yes, you're exactly right. It is a soap opera. It's been distributed to over 1,100 universities, and actually high schools, too, as 28-minute episodes that are shown every week as supplementary or collateral material to Spanish instruction.
HS: On your screen, do you see this as a full-screen thing or do you see it in one of these little two inch boxes?
BT: It's not a two inch box, it's generally about a quarter inch size would be its natural resolution. But students can also -- MPEG1 video can actually blow up to full screen size. It gets a little bit blocky, but you have 30 frames per second. You have very good audio associated with it, and it's basically sort of like being able to play a VHS tape on your computer screen.
We found that the faculty members, at this level of quality, said, "I want to start using it!" The students took to it and we learned a lot of things about the challenges of delivering video to 100 students on campus in real classes where the video needs to change from week to week with this experiment.
It was basically a success, but we also learned things that we had to deal with. For example, even though we were able to serve this video seven days a week, 24 hours a day, for some reason the students still chose to wait until the night before the assignment was due to watch the video.
HS: That would be our talk on psychology.
BT: We had very --
JB: Right, human behavior!
BT: But I haven't solved that problem as to how to spread it out.
HS: You mentioned MPEG1 and I've heard of MPEG1 and MPEG2 and even MPEG3 (which I think doesn't exist) and MPEG4. Could you tell us about all these MPEGs, what they are and what they do?
BT: Yeah, I think it's really important with people -- with universities moving into using digital video -- that they pay a lot of attention, because this really is going to be an institutional planning effort to put in place the network and to protect your investment in your content. I really want to encourage people to look at the standards that are out there and that make a tremendous amount of sense to me and that have been adopted by the broadcast industry, by the industry that's producing DVD ROMs.
And a lot of these standards have the surname MPEG. MPEG1 is, I think, the one that is the standard of greatest interest to universities today because it's good enough quality that it will attract the favor of your faculty members. And it really delivers very high quality video -- certainly good enough quality, but its bandwidth intensity is about one and a half megabits per second. That means that if you can deliver a switched ten Ethernet connection to a desktop machine -- and I want to emphasize switched --
HS: Switched, not shared.
BT: -- not shared Ethernet -- you can deliver great quality MPEG1 video. And I think that's kind of a sweet spot for universities in terms of both quality and the networking infrastructure that is within their grasp.
HS: Yeah, though most universities, I think, are using shared ten megabit.
BT: Based on early applications that we did in a couple different disciplines here during 1998 and 1999 at Northwestern, we showed enough potential and enough adoption of using MPEG stream video that the university made a decision a few months ago to, this coming summer, upgrade all 5,000 residence hall ports at Northwestern from shared Ethernet to switched Ethernet.
HS: That's great.
BT: This was not driven by kind of some technological, oh, priority, that for some vague reason we should be at switched Ethernet. It was only hearing from faculty members that they would use this in American studies, in foreign language, in materials science and in chemistry that the university said, "Now is the right time to invest." In the case of Northwestern, it's going to be about a half-million dollars, from shared Ethernet to switched Ethernet, as really the student computing environment here.
HS: And you said that MPEG1 uses one and a half megabits.
BT: One and a half megabits per second.
HS: Could you give me some idea of how big a file this thing is? I mean, if I have 30 minutes of MPEG video, how big is that thing?
BT: Joel, can you do the calculations real quick on that? One and a half megabits per second times 60 times 30.
JB: Times 28 minutes, right.
BT: Yeah. Yeah. It's really manageable with the size of storage, with the price of storage now on video servers.
JM: Yeah, storage is cheap these days. So that's really not a major issue. But let me mention something about MPEG1 that I think is a major issue, which is that one of our objectives in these projects is to make digital video ubiquitous. That is, we want everyone to have access to these resources.
And as a consequence, we've standardized to some degree on MPEG1, 1.5 megabits per second, for many activities because you can currently buy an off-the-shelf PC and do decoding in software in order to see those streams. In other words, you can see full motion, full color, full screen video without having to buy an extra board for your PC.
BT: Any MAC or any PC that you've bought within the last year can decode MPEG video.
JM: Exactly as it should be.
HS: When you say MPEG, you mean all the MPEGs?
BT: No, only MPEG1.
JM: Just MPEG1.
BT: If you go into MPEG2, which is the quality of video that you see on DVD ROMs now or if you have a direct broadcast system at your home with one of these 18-inch satellites, that entire industry is centered upon MPEG2 video. It's a larger screen, it's a higher refresh rate, and basically you need to figure on three to ten megabits per second for typical MPEG2 video delivery.
HS: Okay, so your ten megabit switched would still work.
JM: No.
BT: At the low end, and you aren't going to be able to do it with just your standard MAC or PC workstation without adding on a hardware decoding card.
JB: What is the software that's on the basic MAC or PC right now that's doing this decoding?
JM: It's a video client with some Java plug-ins for most people. In other words, a fairly free client and this is another objective to making this ubiquitous.
But just to finish up on the MPEG1 and the ubiquity of it. If administrators were faced with the idea of having to put decoding cards in several thousand machines on their campus, they simply wouldn't do it. So this is another reason to think about MPEG1.
However, with regard to the MPEG2 that Bob was describing, soon there will be PCs powerful enough and software clever enough to do decoding and software for MPEG2. We expect that to start appearing sometime this year, so over the longer term, you'll see these ultra-bright, ultra-sharp pictures that Bob was describing -- MPEG2 -- on PCs.
HS: But then you're saying that ten megabit switched is not going to be enough.
JM: That's a different issue.
HS: Okay, we'll have to go to hundred megabit switched?
JM: That's not the PC issue, that's the pipe to the PC issue.
HS: Okay, but, you know, this is, we have a car but we don't have any roads to drive it on.
JM: However, if MPEG2 can be streamed at anywhere from three to ten, as Bob said, megabits per second and it is good enough for the lower rates.
BT: Yeah. I think that MPEG1 is the great place to start if you're determined to be working in and developing content in high quality video.
Continue to watch the scene with MPEG2, but you're going to have a lot of work to do just to take your next steps in your campus network infrastructure deployment to make it safe and secure for MPEG1 video to be streamed to all of your instructional and kind of academic computing locations.
HS: Okay, could we talk just a bit about MPEG4?
BT: Yeah. You've skipped MPEG3!
HS: Tell me about MPEG3 because I can't find --
BT: You asked about that the other day and I looked this up. There actually was an MPEG3 committee and it turned out, MPEG3 was originally designed to develop the standard for HDTV. And it turns out, they found out that they could continue to raise the bandwidth rate of MPEG2 so they disbanded themselves or were terminated. And so there's no longer an active MPEG3 committee.
HS: We should mention that MP3 is not MPEG3.
BT: Correct!
HS: So no one out there should get them confused!
BT: The other MPEG committee that's very active here now is MPEG4. And MPEG4 actually was originally convened (the committee) to develop very low quality video for video phones. You know these commercials we've been seeing since the 1950's for how everybody's going to want to have their phone have a camera and a screen on it, and that market has gone nowhere.
Well, anyway, MPEG4 was originally developed or convened to address video phone type video, but they soon realized this was not going anywhere. Instead, they turned their attention to video that really anticipates the entirely digital creation and repurposing of video. It's video that is not passive video, the way MPEG1 and MPEG2 or other video streams are typically being used on campus now, which is much like a tape player machine: you start the video at the beginning of your clip and you kind of play it to the end, and you can't do much other than watch it and maybe kind of scroll back to the beginning or to the middle.
But in MPEG4, they're breaking up video into modular objects and you can combine graphics, live video, stored video and the video can even respond to either cookies or profiles that it detects on the client to change the appearance or different aspects of the video stream for the individual user.
HS: What kind of quality is that going to be? Is that going to be HDTV quality?
BT: Originally, it was targeted to be low quality video, and I've been watching the MPEG4 proceedings. I'm a little confused by what they're saying. They say that they are optimizing MPEG4 for low, medium and high quality video, and that sounds a little suspicious, doesn't it?
JB: Sounds like they're trying to do everything.
JB: At this point right now, Bob, what's the sense of the bandwidth that will be needed for MPEG4?
BT: In general, what we're seeing with MPEG4 is low to moderate bandwidth, and by that I mean anything from a 56K modem to 500 kilobits per second. QuickTime 4.0 is probably the closest thing that's out there in wide circulation at the moment to what MPEG4 is becoming.
HS: But does QuickTime 4.0 use MPEG4, or are they two different things?
BT: That's a tough question. MPEG4 is not done being finalized, and the MPEG4 committee has decided that the QuickTime file format is going to be the file format that MPEG4 is going to be using.
That is not the same, though, as saying whatever you see in QuickTime 4.0 is the same, identical and fully compliant with MPEG4. We're watching these technologies develop, and the MPEG4 standard develop, and we'll continue to watch and sort of see whether QuickTime 4.0 becomes synonymous with MPEG4 or whether it's going to be a variant.
HS: Okay, we have a few questions that have come in.
JM: Before we get onto these questions, I have a few more comments about these standards and these responses to frequently asked questions.
One is, if I encode a lot of material in MPEG1, is that going to be obsolete over time? The general answer is no because there will always be people at low bandwidth pipes that could use that material.
Another question is, what is the use of MPEG2 today? And generally, that's in distance education where you have dedicated classrooms that have the de-encoder boards installed. That is, there's a program that's set up to deliver high quality education to specialized audiences and those types of purposes are good for MPEG2. You're not trying to distribute over a whole campus; you're doing dedicated, specific work. Also it's good for such things as medical imaging classes where you need very high resolution.
And then finally, people want to know when the MPEG2 decoding software will be available, and again, maybe third quarter this year.
JB: Okay, good. Going back just very quickly to your one point, Joel, about the MPEG1, in conjunction with one of the steps that Bob has outlined, i.e., acquiring video. Are you doing any encoding at a higher resolution, and then keeping that as source so that when these other higher formats are easily deliverable that you'll have that? Or are you just going to go ahead and do the MPEG1 encoding?
JM: That question comes up a lot, which is that if you are doing encoding, should you just do it in MPEG1 and MPEG2 at the same time so you don't have to go back to it. And generally, I think that's a good idea if you can do it.
HS: But then what's going to happen when people start using MPEG4?
JM: Well, that's a different format. Well, that's sort of a different purpose, actually.
JB: Okay, so the content, in fact, may be different.
JM: Yes, it's more for interactive than, for example, for store and forward, for streaming.
HS: Okay. We have a few electronic mail questions here. There's a few from Stephen Davies at [inaudible] College in Calgary. Stephen says, "I attended a demonstration of video streaming last fall, but the Apple presenter complained he had to use an MPEG demo so his product could be properly compared with that of the other presenters." Obviously, he would have been better to demonstrate in QuickTime. And Stephen's question is, "How much is competition holding back the use of better technologies in this developing field?"
JM: Well, I think that's an excellent question because one of the problems right now with video is that almost all of the products that are generally available on the market are closed, proprietary systems, and I think what really is needed is something that is much more open.
It becomes a practical concern for colleges and universities because what they would like to do is simply give out a video client that would be able to read a lot of the formats. The problem today is that everyone has to have on their desktop ten different clients for reading ten different proprietary formats, and that becomes a difficult situation to administer for the universities -- and it becomes very difficult for those people who are trying to incorporate this technology into classes and into research. So that is a problem and I think that at some point it's going to be solved with a more open type client. That is, things can be proprietary, but they ought to be inter-operational.
BT: Howard, the e-mail that you got is correct, though, in that there is a lot of both competition and sharp elbows amongst the various players who are developing video technologies as to how interoperable they decide their particular system and player and approach should be. A lot of the activity that you see right now in the marketplace is focused upon Web based video -- lower quality because of lower bandwidth video -- and that's where we're seeing more of the proprietary encoding formats being marketed and appearing on people's desktop machines.
My plea, when I'm working with the different friends we have in the industry, is to have them not only be engaged with that proprietary lower end, lower bandwidth market, but also to include the MPEG1s, the MPEG2s and anticipate the MPEG4 in their products. And I'd like to see more of that done by the industry players.
HS: Stephen has a related point here. He says, "I'm afraid to download Windows Media Player in case it assumes control of any other media activities on my desktop. Is there a simple way to know we are keeping our options open when we download these files?"
BT: I think that's going into a question beyond the focus of today's TechTalk.
JM: Right!
BT: But actually, I'll give Microsoft credit for -- they are very involved with the MPEG4 committee and development effort and they're --
HS: I've seen a lot of stuff about MPEG4 out on the Microsoft site.
BT: Yes, and their Netshow Theater, which is their kind of premier video server product, actually supports a wide variety, and I think a nice suite of encoding formats.
HS: Okay, Stephen has one last question here. Stephen, you're probably sending more questions right now, but this is your last question! But Stephen says, "I've seen demonstrations of digital video editing, but the highest precision still seems to be second by second, even though there's many frames per second in the streaming. Is there any sign of this improving soon?"
JM: Why don't you mention the Avid work you're doing, Bob?
BT: Here at Northwestern, we have set up a dedicated work group called the Northwestern Advanced Media Production Studio to focus mostly upon digital video creation and figuring out how to move this quickly from acquiring content to being able to use it on our network, in our classrooms and with our students.
I'm finding that it's only good news in terms of what you can choose from in terms of editing workstations to edit digital video. The Apple has certainly raised the consciousness of people here with their IMAC DV and their G4. And the price of Avid systems -- and Avid is one of our partners in our advanced media production studio -- has come down from $100,000 or $200,000 to machines that are now being offered at about a $10,000 price range that it's really just amazing what they can do now with DV.
And by that, I'm referring to an industry standard. DV video can now be done on a five and ten thousand dollar machine with amazing results, stuff that you just didn't think you could do two years ago on anything under $100,000. And certainly, those machines such as the Avid DV Express, which is a new product that came out last month in cooperation with IBM, sells for about $10,000 -- it's doing more than second-by-second frames. I'm sure that's going up to 30 frames per second at least.
HS: Okay. We have another question asked here from Richard Bloom from University of Pennsylvania. And he says he would be interested in knowing what the experts' views are on the best format for streaming media for modem users. He has passed along to us a URL which Terry has picked up on a Streaming Shootout Website out there, and Terry has it so it'll probably be posted on our Website. But could you talk -- he says that the shootout involved looking at Windows vs. QuickTime vs. Real vs. all these things.
BT: At Northwestern, we're intentionally focusing on the high quality MPEG video. Nevertheless, I know from what we see happening on this campus and others that you have video servers from REL, from Apple with their QuickTime streaming and from Sorenson with their Sorenson Broadcaster, all of which do acceptable quality as long as you're thinking of Web based video.
But you're really talking about one frame per second, three frames per second, at most five frames per second even under the best modem conditions with that type of video. In that type of video, the lower bandwidth, Web based video, the attention is being paid, and there's good reasons for it, for maintaining the audio stream always at the expense of the video stream.
HS: Okay, we actually have another question. Actually, the questions are pouring in here.
JB: I was just going to ask Howard to remind folks to send in questions. Maybe I shouldn't do that!
HS: There's no reason to do that! Folks know, those of them with questions. Okay, and there's Mike Heidt from PSU. Where's PSU?
BT: Penn State, yes.
HS: Okay, that's a university.
JB: I think it's Portland State.
HS: Okay --
BT: No, that's Penn State. We know Mike.
HS: You know Mike?
BT: This may be a canned question.
HS: Okay, Mike, we're going to answer a more general question than the one you asked here. Mike basically is asking a question about broadcasting vs. narrowcasting or unicasting vs. multicasting and perhaps we could just talk about the difference between broadcasting and narrowcasting and unicasting vs. multicasting, and that should include the answer to the question that Mike was asking.
BT: Okay, Joel, do you want to take the broadcasting and narrowcasting and what you're doing with your video portals?
JM: Sure, and I'm sure that Mike Heidt actually knows the answers to all these questions --
HS: He's just seeing if you know them.
JM: --because he's one of the foremost experts in digital video. He's a great guy.
Okay, broadcasting is an interesting concept because it goes back to the early days of TV when there were licensed channels, as there are today. And basically, the resource is treated as a scarce resource to be carefully rationed. And I think we're moving away from that model. You see that with the proliferation of cable channels where there are numerous cable channels now with more targeted types of programming. However, in general, though, with both broadcasting and cable channeling, you still get lowest common denominator type programming.
The great promise of digital video over the Internet is very narrowcasting so that you can do specialized programming for specialized targeted audiences. And I would understand that Penn State would have particular interests there because they excel in many areas of distance education, and this allows them to have customized programs for custom audiences.
But in addition, for just general types of channels, for example, let's say that you don't want just a history channel and not just a Roman history channel, but you want an early Roman history channel related to southern Spain. Well, the Internet allows you to do that. It is much more flexible as a medium.
In addition it's flexible in the types of programs you can put on those channels. For example, you're not narrowed down to a 30 minute segment vs. an hour segment. Within the Internet, you can have programs that are two minutes long or two days long (if you care to see that much video!). So there's much more flexibility, much more capability of targeting.
And then you have a reach that's actually worldwide, which is another differentiator. With broadcast, you reach a particular geographic area, but over the Internet you can create communities interested in similar subjects that are worldwide. So in one region, there might be only ten people, but worldwide, there's a couple thousand that want to see particular kinds of programming.
HS: Do you have this same kind of distinction inside a university? Could you talk about broadcasting inside a university vs. narrowcasting within a university.
JM: Well, of course, and Bob gave an example of that earlier with Destinos. Destinos would be broadcast to the Spanish classes. However, everyone may wish to see the broadcast of the university delivering the annual state of the university address, which may be more widely seen.
Or something like the Mars global explorer broadcast over the Internet2 to all of the people who would actually like to see the live casting. One project we're doing here is we're taking NASA TV and we're putting it on the Internet so we have a transmission of NASA TV. There's some very interesting programs there that a lot of people are interested in, for example, when the Challenger is launched, you get a higher percentage of people wanting to watch that.
BT: Howard, you also mentioned two other terms, unicasting and multicasting.
HS: Yes.
BT: Let me take a shot at this because it's a -- well, it's part of your bandwidth brokerage calculations as you begin to support digital video on your campus. You're going to have to strike a balance between what's called unicasting and multicasting.
Unicasting is one-to-one between the receiver of a video stream and an actual video stream having to be launched by the video server and sustained over the campus network. Another name for it is often "video on demand". This is often the method that many of us get involved with with first introducing high quality digital video to campus, and that's what we did with Destinos is --
HS: But do they go off to a video server and they say, "Right now, I want to watch this"?
BT: Yes. Doesn't matter if it's 6:32 in the morning or 7:54 at night, that resource is always available. But it's always one whole MPEG stream that's having to be carried over the campus network to serve one individual.
HS: Okay, and if somebody wants to watch the same thing but they want to watch it ten minutes later, they're looking at a different video stream?
BT: Yeah. So very soon, you could -- if you're supporting Spanish or general chemistry class with MPEG video -- you could see having too many video streams being requested to sustain over your campus network.
HS: Right, even though we're all watching the same thing, we could be watching at different points.
BT: Go ahead, Judith.
JB: Yeah, let me just make a comment here because this unicasting then is basically what we're doing on an audio level.
BT: Yes.
JB: Because everyone calling in for this Webcast, really, each one is receiving a different stream.
BT: Yeah, and that hasn't become a problem yet because audio streams are pretty small.
But when this happens with video, what you need to do is programmatically begin to decide what resources instead of unicasting or video on demand you're going to move over to multicasting. And what that means is that you announce to the campus (and you probably have a Web page up to show the schedule) that you're going to stream Destinos every 30 minutes throughout the day. But this is many people being able to take advantage of one video stream, and this conserves your network bandwidth. And if you do it right, it doesn't impose unacceptable standards or requests upon your users. If they can get to the 30 minute episode, you know, on the hour and the half hour, that's probably good enough and it'll save you a lot of network bandwidth.
There'll be other types of resources. We've developed 83 hours of video in an archive called the Video Encyclopedia of the 20th Century. And it wouldn't be acceptable to just decide we're going to stream different clips of the video encyclopedia at designated hours on campus. This is a database and query drive video archive, and for that, all of the video coming out of the Video Encyclopedia of the 20th Century is video on demand. But Spanish instruction, maybe chemistry lessons, can all be multicast at scheduled times and it's much more economical use of the network.
JM: And now for our good friend Dr. Bob Dixon at Ohio State and at OARNET, we're going to have to say a few words about videoconferencing.
HS: Okay, why don't you say a few words about videoconferencing? We'll pretend we got electronic mail from Professor Dixon.
JM: Yes. I'm sure we may have!
Well, the broader context is that one of the modalities beyond live transmission (which we've discussed a bit) and beyond video on demand (which we have also discussed), there is the intriguing area of videoconferencing, which is particularly important to distance education, but it's also important to researchers who are collaborating and for, even to some degree, students, either for social purposes or for working on joint projects. So it's an important modality.
Currently, there is much work being done on a protocol called H323, which is a protocol which allows for videoconferencing over the Internet. And you can see why this is a natural, because people use the Internet primarily for communication. Now, often this is one-way e-mail or interacting through e-mail. But with the medium sitting on the desk and with a tube and a little camera, you can interact with people and it's quite easy. You just fire up a client and tune them in.
There are technical challenges to be resolved and Bob Dixon is one of those people working on some of the intriguing work in this area. One of the things that Bob has done recently is he organized this thing called the Megaconference, which he did for the Internet2 conference in Seattle last fall. He worked with a bunch of people in partnership to tie together almost 50 organizations -- not only throughout this country, but also throughout the world -- who are linked together through a device called an MCU so that 50 organizations throughout the world could be linked together and participate in the same videoconference. And what he did was he organized this so that each organization would announce itself and talk a little bit about their video projects and then the baton, so to speak, would be passed to the next person.
The interesting thing about that was that people, before that happened, said it could not be done. There were vendors of equipment that said equipment couldn't do it, but there it was! And it shows you that you can do large scale type of communications through this intriguing technology, not just the small two, three, four, five people together discussing. So it has wonderful promise.
Some of the missing pieces include such things as a numbering system or an addressing system. For example, everybody has a telephone number and everybody knows what it is and how to use it. You need something equivalent in the 323 world and that hasn't quite emerged yet, but that's another area that Bob Dixon is working on.
HS: Okay, strangely enough, we are getting close to the end of this Webcast, so we've got to try to get our last few questions in here. There's actually several e-mail questions. I'm just going to do one of them here. For the other ones, we'll answer them. After the broadcast, you'll see them in the archives. But Tom Horton at Cornell wants to know -- he points out that these digital video streams really gobble up a lot of bandwidth and he wonders what's going on to solve this problem.
JM: Well, the main thing that's going on is at the national level -- the Internet2 project which is providing for a national backbone for 160 universities through a collaborative effort.
And then in addition, there's certain techniques that are being developed to optimize that bandwidth. One is the multicast that Bob mentioned. Another one is a project called Quality of Service in the Internet.
Also, I'm going to use this opportunity to put in a plug for the Internet2 Digital Video Project, which is a consortium under the auspices of Internet2 which is bringing together groups from NizerNet, which may include Cornell (I think it does), Research TV out on the west coast, the CIC MREN people in the Midwest, VDAY which is southwest but now going national.
So there's many people collaborating throughout the nation on different kinds of projects to be able to enhance the environment for everybody. And so I would like to urge everybody to get involved in these projects if they can.
JB: Well, I think that we have the link to the Internet2 Digital Video Web page up on the site, so --
JM: Great.
JB: People can go ahead and link to that. One thing that you didn't mention that I kind of thought you might in terms of responding to the question of managing this bandwidth is what about compression techniques? Is there any new development kind of on the horizon in that area?
BT: There continues to be a tremendous amount of work. Joel, can you talk about either the Wavelets or the others?
JM: Yeah, there's Wavelets, there's Fractal Compression, so there's a lot of good research there where the essential direction is better and better compression with less and less loss of data. So there's much work being done there.
And of course, that's important. What's more important, though, as a technology for optimizing bandwidth (which relates directly to the questioner's issue) is replication services, and I think that actually has --
HS: What are replication services?
JB: Yeah, right.
JM: Well, imagine, if you will, there being a very important archive that people would like to see of a certain kind of material. The best way to actually present that to people might not be one big server sitting in Kansas, but rather to take those materials and to distribute them --
HS: So like a bunch of mirrored sites?
JM: Exactly. The difference being, though, that it's not just replicating materials sometimes, but it's also replicating the access methods for those materials and certain kinds of related files, such as those having to do with property rights and security and priorities and those types of things.
HS: Okay, for listeners or people who are going to look at this archive later, if one wanted to get started in this area, what are the first couple steps that people should be taking?
JM: Which of the many areas we discussed?
HS: Well, you said that folks were going to be distributing digital video stuff on campus, you said that folks had to acquire, edit, encode and distribute this stuff. I mean, if I wanted to take something like Spanish soap operas and I wanted to distribute these things on campus, how do I get going?
BT: Well, the first thing you're going to do is check your copyright situation.
HS: We could do a whole TechTalk -- in fact, we have done a whole TechTalk on copyrights and intellectual property.
BT: Yeah, and actually, Judith and CREN are --
HS: So the first thing to do is go back and look at the archive on that.
BT: Yeah. I like to break down the general process of both planning for and actually executing video projects on campus as being divided into those four areas that you mentioned Howard, which is you need to figure out how you're going to acquire the video, how you're going to edit it. You don't edit it if it's videoconferencing, but a lot of projects on campus involve an editing stage. How you're going to encode it, and there you're making calculations as to what type of bandwidth.
HS: This is the MPEG1, MPEG2 decision.
BT: MPEG1, MPEG2, QuickTime 4, or if you're doing a demonstration project, you might before you know it -- it might happen sooner than you think that you might be doing HDTV. And then the fourth step is figuring out how you're going to distribute it, which to me means usually streaming it.
In terms of acquiring it, there's a couple technologies I think people should be looking at when they go out to put together systems. One is DV. DV is the standard that's been developed and supported by over 50 companies out there. When you go to buy a camcorder or a digital camera now or other peripheral device, look for DV, and that's Digital Video format for not only the video format but helping to assure easy transferability between your different peripheral devices for digital video.
The best way to move stuff between devices, say, between a camcorder and an IMAC or a PC workstation is with FireWire. That's another technology you should be aware of. The formal name for this is IEEE 1394. That's the standard committee.
HS: Oh, that's much more catchy than FireWire!
BT: Yeah! I like it, but Apple calls it FireWire, Sony calls it I-Link. What you should know is that they're all the same thing, which is a serial but high speed transfer protocol that really is the way to go right now for connecting a camcorder directly to your PC.
In terms of editing, you look at non-linear editing stations. They used to cost you $20,000, $50,000, $100,000. Now you can look as low as $2,000 for kind of home systems to $5, $10 and $15K for good non-linear editing systems. And you want to make sure they can work with DV, that digital video standard I mentioned before.
Encoding, I suggest that people do the highest quality encoding that they can support on their campuses because, in my case at least, the faculty members gave me a very clear signal that they were interested in high quality video but not in the lower quality stuff.
Then in terms of distributing it, you have a number of video servers to use. In terms of enterprise level industry strength video servers, here at Northwestern and many of the other Big Ten universities, we're using the IBM Video Charger, but there's also Sorenson Broadcaster, there's QuickTime Streaming, there's Real, so you have a lot to choose from.
But by dividing your planning process and your actual development process into figuring out how you're going to acquire, edit, encode and distribute, I think you'll make life a lot easier.
HS: Yeah, I was looking at the first letters of those things to see if we can come up with --
BT: I'm looking for an acronym, Howard, I figured you would come up with by the end of this.
HS: I need more time! We'll mention it at the next TechTalk.
JB: That's right.
JM: I'd like to mention a couple other sources of information. One is the VIDE site which you can get to off the Internet2 Digital Video site. It has a couple of white papers that they have done on digital video and it has a lot of good background information. They did a couple of digital video cookbooks which are very good.
JB: Okay, great. I think we're probably at the time when I generally ask, Joel and Bob, do you have any final comment that you'd like to make before we wrap this up?
BT: Judith, this is kind of a technology development that I think people could spend the next ten or 20 years of their career at universities in. So I suggest that people take the long view, look for how they're going to develop and distribute as high a quality video on their campuses as they can right now, and this technology is moving very, very quickly.
JB: And given that, I think you and I talk about us having another session on this in the fall, focusing on instructional applications.
HS: Hey, we'll make you commit to this right on the air!
JB: That's what I was trying to do!
BT: I said I'd help organize this thing!
JB: Okay, Joel, final comment?
JM: Well, I would just like to thank you for the opportunity to talk about this exciting new subject and then also to remind folks that when the Web first came along, there was a lot of puzzling questions about it. Would it fit in a university environment, was it needed? And so people exhibited a little bit of skepticism about it.
In the same way, there's a little bit of that skepticism about digital video on the Internet, but there's a very close parallel between the adaptation requirements and processes for digital video. You're going to see this, as Bob said, move very, very fast. And I would suggest that everybody keep an eye on this and get involved with what projects they can.
JB: All right, thank you. And Howard?
HS: I'm afraid to ask another question because we'll be here until 6:00.
JB: Almost!
HS: But we do have lots of other things we'd like to talk about and we do hope that we'll be able to get you folks back and do some more. This is all very interesting stuff.
BT: It'll be our pleasure, and in three months, we'll have lots of new interesting things to report. Sounds great.
JB: We only got through half of our questions starting out, you know. I'm sure we will have a lot to talk about. So for now, thanks to everyone for being here today and you can go ahead and send follow up questions to expert@cren.net and we do answer those as best we can.
Many thanks to all the institutions who help support these TechTalks, and we invite you and your institution to help support these TechTalks if you're not already a CREN member. Be sure to print out the calendar from the Web page for this spring. Be sure to set aside time on your calendar for another session two weeks from today, on February 17th, for a conversation with Ken Klingenstein and Keith Hazelton with an update on what is happening with campus-wide directories and authentication design and software.
Thanks to everyone else who helped make this event possible today and to our guest experts, Bob Taylor and Joel Mambretti; to technology anchor, Howard Strauss; to Terry Calhoun, our event page producer; to David Smith and Patty Gaul of CREN; to Julia O'Brien, Jason Russell, Carol Wadsworth and the whole support team at the MERIT Network; to Susie Berneis, audio file transcriber; Laurel Erickson, transcript editor and indexer. And finally a thanks again to all of you for being here. You were here because it's time.
Good-bye, Bob and Joel. Bye, Howard.
HS: Bye, Judith, bye, Bob, bye, Joel.
JB: Take care, we'll see you all soon.
HS: Right, bye-bye.
JB: Bye.
