LifeSize has made their announcement.

[Gary Miyakawa]

Wainhouse did a special on it...

http://www.wainhouse.com/files/wrb-06/WRB-0614.pdf

The first of the truely HD solutions....

Let the games begin !

Gary Miyakawa

[mazzarak]

These products look very impressive at first glance - the price isn't too shabby either...!

[Sean Lessman]

Wainhouse did a special on it...

http://www.wainhouse.com/files/wrb-06/WRB-0614.pdf

The first of the truely HD solutions....

Let the games begin !

Gary Miyakawa
Two questions:

1. What bandwidth is required for 'HD resolution'
2. Is this using H.323/H.264 or some proprietary algorithm? Wainhouse didnt specify.

[cribbinsb]

1) From the wainhouse article:
"The codec system connects to any display and delivers up to 1280x720 resolution at 30 frames/sec with a 1Mbps connection. At lower bandwidths, the codec maintains frame rate but drops the spatial resolution."

2) The datasheets are now up on www.lifesize.com. It is using standards based H.323 or SIP with H.264 video, and Mpeg-4 AAC audio.

[Sean Lessman]

I find it interesting that on announcement day, there are 'new' posters on both VTCtalk and Wainhouse pumping hard. I hope Lifesize has a bigger marketing budget than that :-)

I look forward to taking a peek, as on paper, I do not see anything that is revolutionary, but I suppose I could have missed something. H.264 makes allowances for HD encoding. H.239 has been done by everyone and AAC has been readily adopted. The phone interface has been done by Polycom, although the Polycom phone doesnt do 22kHz.

Whats the plan for infrastructure? MCUs, gatekeepers, firewall traversal? Right now, a few endpoints, gateway and management software limits the targets.

I also cannot think of too many people doing 1Mb video, and certainly do not think customers will raise the usable bandwidth to 1Mb to use HD. I would also imagine that 1Mb is the minimum for HD and probably requires closer to the 5Mb to achieve the quality picture described in the announcement. You can do 30fps at 56kbps, but its not going to look good.

I guess I will need to see it in person to appreciate the announcment.

Any ideas on technology under the hood?

Sean

[Entropy3XD]

The price and 8-way MCU as standard is revolutionary enough for me, although I have yet to take one for a test drive to see if one gets what they pay for.


Commenting on the marketing budget is a bit premature on the first day. If you were a part of a new company hitting the market to go against "The Big Guys", then you would probably be just as excited to get the word out in your own way, especially if you have had to remain tight-lipped for some time.


As for infrastructure, didn’t Tandberg just start out with a few endpoints, with the MCUs, GKs, and firewall traversal only coming into serious play over the last couple of years. We just recently got serial MCU functionality with the MPS (don't even mention the back to back Adtrans, it functioned but it was a PITA). Tandberg and Polycom have an outstanding product line, but it took many years to get there.


I agree that using 1MB for HD won’t be feasible for many, but all things being equal and forgetting about HD, I would like to see these units head to head with competitors at 384k and see who has the best quality. I would also like to know what is under the hood and I look forward to some competition between the manufacturers. After all, competition can breed some amazing results.

[Sean Lessman]

Commenting on the marketing budget is a bit premature on the first day. If you were a part of a new company hitting the market to go against "The Big Guys", then you would probably be just as excited to get the word out in your own way, especially if you have had to remain tight-lipped for some time.


As for infrastructure, didn’t Tandberg just start out with a few endpoints, with the MCUs, GKs, and firewall traversal only coming into serious play over the last couple of years. We just recently got serial MCU functionality with the MPS (don't even mention the back to back Adtrans, it functioned but it was a PITA). Tandberg and Polycom have an outstanding product line, but it took many years to get there.


I agree that using 1MB for HD won’t be feasible for many, but all things being equal and forgetting about HD, I would like to see these units head to head with competitors at 384k and see who has the best quality. I would also like to know what is under the hood and I look forward to some competition between the manufacturers. After all, competition can breed some amazing results.The marketing budget comment was my attempt at poking fun.

I also look forward to seeing the products and welcome the addition to the market. Hopefully the additional players will encourage growth for the whole industry. Wouldn't that be a wonderful thing!?

Lastly I will say that the catalyst for this market is not price and video quality. If so wouldnt the revolution already taken place with the cheaper products and H.264 available today?

Sean

[Entropy3XD]

Lastly I will say that the catalyst for this market is not price and video quality. If so wouldnt the revolution already taken place with the cheaper products and H.264 available today?

Sean
Very true. I myself will always be willing to pay extra for a product that is highly adaptable and durable for different environments, from elegant conference rooms to the indestructable deployables. This is one of the reasons I have always been a Tandberg fan.

From what I have seen so far, the LifeSize has most of what I would want in a codec at a very attractive price, assuming everything works. Will it fit all environments? I guess we will have to see. Once everyone gets a chance to test drive the new products, I am sure we will find areas of improvement. There are always areas of improvement. I can’t wait to put one in a Hardigg or Pelican and take it for a spin in the back of my truck to test the durability http://www.vtctalk.com/images/smilies/biggrin.gif.

[Narayanan]

HD resolution may not really be that attractive to the customers even in applications like telemedicine, as bandwidth cost is a constraint, especially in India. Improvement on H.264 would be welcome.

The camera is not movable, is that so?

[Skylark]

Is it compatible with other VC equipment?

[Entropy3XD]

It is standards based, so it should be compatible with other standards based equipment, but we all know how that can go sometimes.

[Sean Lessman]

Is it compatible with other VC equipment?
Should be. I guess we will find out. If not, that will be a shame.

Sean

[Skylark]

Thanks, I read through the info pretty fast and I didn't see anythinga thast said it was standard based or H323 compatible or anything like that so I thought I should ask.

[trapehzoid]

8 way MCU? ipower does more doesn't it? Didn't mean much..

HD resolution? can make up any resolution I want doesn't mean its going to look good. If its at multiple mbit speeds.. and your primary concern is image quality.. why bother with H.323 anyways.. Goto a MPEG encoder or something like a VBrick. Compatibility isn't going to be a problem, because at those speeds.. talking to the masses isn't your concern anyways.

Look at systems that do XGA today.. no one can do it at more then 10fps. So now you are going to increase the resolution by 4x and expect someone to 'magically' overnight (remember.. everyone is using off the shelf DSPs) increase performance over 12x? Remember.. if it sounds too good to be true.. it is.

Even if you tell me they aren't using standard DSPs.. I find it hard to believe they have the capital or time to engineer a brand new silicon at those performance levels.

Put these things together and that draws me to the conclusion they are simply doing custom resolutions in H264 (which everyone can do) and using bandwidth as an excuse for compression ability. Image quality at 5mbps is meaningful for a handful of people (there are people who try to use H323 as a replacement technology for analog systems, distribution, etc), but certainly not the technology that is going to rock the industry or scare anyone.

The most interesting thing about it IMO is trying to commotitize a HD PTZ camera. Is everyone here jumping at the price point willing to buy a HD display ($$$) to make use of it.

The real question is how stable, compatible, and what its image quality is like at 128-512kbs. That's where you seperate the men from the boys.

[Entropy3XD]

HD resolution may not really be that attractive to the customers even in applications like telemedicine, as bandwidth cost is a constraint, especially in India. Improvement on H.264 would be welcome.

The camera is not movable, is that so?
The room system has a PTZ camera, but it looks as if the desktop exec model only has zoom.

[Glen Sykes]

I saw the LifeSize kit today, it's impressive.

H.264 has 2 elements to it, Baseline and Extended (I beleive to be the wording, feel free to correct). The extended element of H.264 accomodates all sorts of stuff, including HD. If you refer to the chart in the Wainhouse article, HD resolutions (1280 x 720) are acheivable at 1Mbps. Additionally, LifeSize are saying they will acheive a resolution of 672 x 384 at 384Kbps. All resolutions stated are at 30Fps (it would be pointless producing the graph at any other frame rate).

Here's a link to some useful info on H.264 Wikipedia H.264 (http://en.wikipedia.org/wiki/H.264)

My feelings on this are that LifeSize, along with that other young upstart Codian, should be welcomed into the VC fold as a breath of fresh air. I for one am sick of hearing the rhetoric of the Tandberg / Polycom axis growing the market, whilst in reality only really serve to beat each other up on features. For example, H.239, the standard that isn't really a standard because it doesn't work between the 2 vendors. It's a joke. AAC vs Siren 14 (now an ITU standard whoopee!), it's not making things easier, it's making them more complex!!

LifeSize (and Codian) have the significant advantage of not having an established user base to keep happy, and therefore are afforded the freedom to innovate without prejudice. My only hope is that Polycom and Tandberg welcome this new competition as a means to help them develop this market, and not try to crowd out the new guys in an attempt to protect their market shares. The VC market, whilst the technology is still exciting, is stagnating from the protective sales policies of Tandberg and Polycom. (Not a pop at you Sean here, you're a great industry ambassador, if only some of the others were the same :D )

I find it quite amusing that there are new accounts posting feverishly about this on wainhouse and here, and it's quite obvious to me where these posts are coming from, but hey, my relationship with Codian started right here on this message board so there's some merit in this.

Trapezhoids typically dour appraisal of the situation is hogwash. Why use Vbrick when you can use something that's standards based, backwardly compatible, competitively priced and is comparable to the emerging standard that people will be experiencing in their homes? For as long as I've been in this industry, my customers have asked for 'broadcast' or 'TV' quality. They have often stated that the picture in their home is better. And consider continuous presence, 4 way split. I can't think of a single customer that is actually happy with the image quality. With HD, a quadrant will have the resolution of 640x360, which is still tons better than CIF. This for me is exciting. HD displays, like all displays are coming down in price. I remember 3 years ago a 50" plasma, with crap image quality used to cost £8,000 (~$14,000). An HD 50" screen is less than this today. My customers bought 50" plasma 3 years ago, no problem. You're right about one thing though, it is H.264, but tell me of any other manufacturer that has an affordable HD camera, a codec with an interface to support it, a chipset with sufficient memory bus and processing power to deliver HD today. What the hell is a 'standard DSP' anyway??? Look at the Trimedia chip (or Nexperia as it is now known), over the last 8 years, it has changed from a 200Mhz clock to a 400Mhz clock. It this in keeping with Moore's law??? Think about it, the broadcast TV industry is moving to H.264 as a means to deliver HD to your home, SOMEONE has to be making hardware for this right?

Based on the premises you have already written Codian's support for H.264 should also be fiction.

I'm sorry to launch into you like this man but you leave yourself wide open to this criticism. I look forward to your reply, but please bear in mind I'm not sure what I can and cannot say about this stuff at present.


To answer a couple of specific questions:

The camera on the room system is a full PTZ camera, with wide angle lens and supports fulll HD resolution. This camera is something else, a really excellent piece of engineering. The exec camera is static with a 70degree field of view. Bear in mind that the Sony EVI-D31 has been around for years now, with little or no development, it's output barely touches a quarter of the resolution of HD. It wouldn't surprise me if the quality of picture from a Lifesize machine were better on a Polycom / Tandberg machine than from a Polycom / Tandberg, just because of this camera.

It is H.323 compatible and also supports SIP. Support for H.320 is provided via the networker device. As with every manufacturer, I expect there to be teething problems with interop, H.323 is a complex beast, however there is a commitment to this working.

By the way, Hello Everyone!

[Sean Lessman]

I think its safe to say that we all welcome newcomers to the market, but some of the points about TANDBERG vs Polycom beating each other up, while appropriate, can also be expanded to include the other entities in the market (partners and customers). If we (the industry) are this excited about one feature, then it shows our maturity as an industry.

I do look forward to seeing the new stuff, unfortunately I will not be at N+I so I wont get a first look. But I hope we (the industry) have higher expectations of our future and expect more than one new feature causing a revolution.

We have made great strides through the ISDN era and are now currently transitioning to IP as an industry. IP offers more challenges and connectivity is more a challenge now than ever. Lots of uncontrollables and many different customized environments out there. I personally see connectivity and reliability more important than a new video resolution. There will be some that find video the most important, there always is, but I don't think video quality is what is holding this industry back.

One last note, that might help with the debate: when we released H.264 to market, most of the questions where around how good it looked at 128, 256 and 384. Nobody wanted to see what it looked like at 768kbps or above. I see the industry progressing towards better video at lower bandwidths. Just my opinion, but I am not the one with the $$$, so my opinion doesn't mean much :-)

We will all just have to watch and see what happens. Right now its all technology demos with no shipping products. Until the facts come out, its going to be hard to have a real debate on the technology.

Sean

[Yukikaze]

I saw the LifeSize kit today, it's impressive.

[snip]
Trapezhoids typically dour appraisal of the situation is hogwash.


I thought Trapezhoid's post was very good actually. I think you missed his main points completely. What he is saying is that:

1. The bandwidth required to deliver the additional resolution is way beyond what is in common use today, meaning you will most likely not be able to interop these features for a long time even if other vendors actually released software upgrades for HD tomorrow.

2. Standard DSPs(meaning the current state of the art system-on-a-chip" designs geared towards multimedia, which is what everybody uses) level the playing field performance-wise and every company that uses them will have a similar level of performance per DSP.

3. Compression standards level the playing field with respect to compression performance. Sure, all implementations of H.264 will not have the same compression performance, but all non-braindead implementations will deliver roughly the same image quality at the same bandwidth.

These three points lead to his conclusion that "If it sounds too good to be true - it is". The sad fact is that if you are standards based there is very little room for high-flying super engineers to do clever things. Because of this LifeSize has done the only thing they were able to do to make waves - add a feature that looks impressive, not because it is an amazing technological breakthrough, but because it ignores real life limitations, in this case practical bandwidth restrictions.


[snip]

And consider continuous presence, 4 way split. I can't think of a single customer that is actually happy with the image quality. With HD, a quadrant will have the resolution of 640x360, which is still tons better than CIF. This for me is exciting.


Everyone agrees that better resolution is a good thing. What escapes you, again, is that the bandwidth required to deliver this resolution is higher than what other vendors allow themselves. Resolution restrictions are there because in real life people aren't willing to pay for the bandwidth required to deliver higher quality.

Your comments about HD in the television world also miss their mark entirely. In addition to comparing apples to oranges(TV sets only decode - Videoconferencing has to encode as well, TV is broadcast and can thus afford the higher bandwith etc) you also seem to be unaware that Nexperia actually is marketed as being able to handle HD. Finding chips capable of encoding and decoding HD in real time is not a problem, finding chips capable of improving the compression ratio over existing H.264 implementations by a large amount while still being standards based is. Even if you lift the real-time restriction.

[Entropy3XD]

Well put Yukikaze.

[trapehzoid]

I saw the LifeSize kit today, it's impressive.

H.264 has 2 elements to it, Baseline and Extended (I beleive to be the wording, feel free to correct). The extended element of H.264 accomodates all sorts of stuff, including HD. If you refer to the chart in the Wainhouse article, HD resolutions (1280 x 720) are acheivable at 1Mbps.

Define 'achievable'. I can send any resolution I want at any frame rate.. doesn't mean its going to look good! I can update as little or as much of the picture I want and with as much detail or not I want. The whole trick to this industry is not who can achieve the highest frame rate or highest resolution, its getting the best picture out of the bandwidth available.

My guess is the 'trick' is they are only updating 1/4 of the screen at most at any given time. Anyone want to guess how long a fast update request is going to take to work at these resolutions?

And I certainly don't consider Wainhouse the 'source' when it comes to anything technical.

The VC market, whilst the technology is still exciting, is stagnating from the protective sales policies of Tandberg and Polycom.

I think its hurt by the 'shiney piece of glass' mentality of everyone demanding useless features and complexity rather then focusing on what would make video reliable, accessiable, and painless to use. The vendors are going to respond to the customer demands. And in general the population sees or hears something new and flashy.. and demands everyone have it.. even if they'll never use it.


Trapezhoids typically dour appraisal of the situation is hogwash. Why use Vbrick when you can use something that's standards based, backwardly compatible, competitively priced and is comparable to the emerging standard that people will be experiencing in their homes?

What does home have anything to do with what we need in the office in this generation? If you haven't noticed.. the lifetime of systems is getting shorter and shorter (typical as technology commoditizes). Why do I give a poo about 'what will be' in 5 years. 5 years from now you'll be buying your next piece of gear to replace what you have now regardless.

Why would I purchase something based on futures if it doesn't do the job I need now? HD ready? Great.. but if it doens't do today's job well, what's the point? I'm not saying the new systems stink at today's uses, but today's use is way more important then the 'potential' future use. What about the quotes in wainhouse today about how bandwidth is getting so much cheaper, 'soon you'll have...' blah blah blah. Hello.. bandwidth has gotten TONS cheaper since the viewstation back in '98. What speeds did we run back then? What speeds do we run now? I think you'll find the two strikenly similar or even LOWER then in 1998.

For as long as I've been in this industry, my customers have asked for 'broadcast' or 'TV' quality.

Asked, but how many of them were willing to pay for it? The quality is available if you are willing to pay for it. Get a MPEG-2/4 system and enjoy incredible quality. Not compatible? So what.. who are you going to talk to remotely or randomly at 3-10 megabits? No one. Those speeds are for closed or private networks.

Also, systems available today can already do 2, 4 and 6 megabits. If image quality was so important to people, why aren't they using these bitrates instead of 192-768kbs? The image quality out of today's systems at those rates is exceptional.. yet how many people use it? A handful. So, please explain to me how lifesize's promise of better quality, IF, you commit to these types of call rates, is going to be widely adopted? People can do it today, with significant gains over using 192kbs-768kbs.. but they don't.. HD promises isn't going to change that.

HD displays, like all displays are coming down in price. I remember 3 years ago a 50" plasma, with crap image quality used to cost £8,000 (~$14,000). An HD 50" screen is less than this today. My customers bought 50" plasma 3 years ago, no problem.

Yup.. note how the marketing hype of Lifesize is all about HD, HD, HD.. but they aren't delivering the resolution most people associate with high quality HD? 1080i. Why? Because those plasmas you like to quote as coming down can't even display those resolutions. They have to downscale it. Thats why Lifesize is touting the 1280 resolution.. because that's the native resolution of most large size plasmas WXGA. It can display 720p, but is only 50% the resolution of 1080i formats. Note the lack of any wording from Lifesize around that? Because they are relying on HD hype.. not HD experience.

You're right about one thing though, it is H.264, but tell me of any other manufacturer that has an affordable HD camera, a codec with an interface to support it, a chipset with sufficient memory bus and processing power to deliver HD today. What the hell is a 'standard DSP' anyway??? Look at the Trimedia chip (or Nexperia as it is now known), over the last 8 years, it has changed from a 200Mhz clock to a 400Mhz clock.

I'm not going to waste my time looking up the performance of the various TriMedia chip generations.. but if you only want to compare the doubling of clockrate as a measure of performance.. don't even bother responding to me. Realize these chips are designed to be in EMBEDDED systems... clockrate = power draw.. power = heat. Both huge negatives in embedded systems. Compare something like the viewstation or tandberg 5000 (both trimedia based) to the systems of today and tell me performance hasn't improved dramatically.

And by standard DSP I mean the simple fact that everyone in the industry is using one of three families of media DSPs. Ti, Equator, or what was TriMedia. 'standard' in that these are available off the shelf, and are used in many different applications besides VTC. What kind of technology do you think is powering those plasmas and other displays???

It this in keeping with Moore's law??? Think about it, the broadcast TV industry is moving to H.264 as a means to deliver HD to your home, SOMEONE has to be making hardware for this right?

Oh my.. if you want to compare the needs of something like a HDTV headend which is broadcasting at 25-100megabits to something that is taking video and compressing it to 1/100th of its original size.. talk about different worlds. There are plenty of people making HDTV broadcast and headend gear. How many of them do you see taking that same technology and using it to deliver high compression, low latency video systems??? NONE.

And you'll note I mentioned their camera as the most interesting development.. before you saw in wainhouse today where Malloy also promoted that as their biggest achievement. I think I have pretty decent ability to see through the fluff and find the meat.

Based on the premises you have already written Codian's support for H.264 should also be fiction.

Huh? what do they have to do with any of this. Codian isn't promising order of magnitude performance increases nor promising things that simply do not compute when you do the math.

It wouldn't surprise me if the quality of picture from a Lifesize machine were better on a Polycom / Tandberg machine than from a Polycom / Tandberg, just because of this camera.

What actually differentiates camera quality in this industry is image noise (as it screws with the encoders) and light handling. Resolution isn't really that much of a deal breaker. You want a higher QUALITY picture, not necessarily a higher resolution input. But again.. ignoring history you shouldl notice how even with 3 chip cameras available, the performance gain has not been worth the cost when dealing with VTC. The VTC industry has been moving towards commodity, not image quality.

[tom9933]

On a related note it looks like Polycom made their HD announcement yesterday. Looks like N+I should be a very interesting show. Now the question is will Sony or Tandberg also show HD conferencing at the show?
http://www.pug.com/news/news.php?cmd=vi&typ=inf&id=87

[mazzarak]

man, I tell you, if you come to this thread fresh, it's a hell of a read. If nothing else, this release (which incidentally I have not had the opportunity to sample) has divided opinion on this forum, which I suppose is a key way of getting publicity. All the more impressive seeing as precious few have actually got their hands on one yet... ;)

Wonder if we have any secret LifeSize presence on the forum here? :D

[Glen Sykes]

Hi All,

Many thanks for the excellent replies, I'm glad I've stirred things up, debate is required if people are to understand the implications of this new entry.

To answer a few points.

With respect to bandwidth, at face value 1Mbps or 2Mpbs, when compared to what the industry has used as its standard for the last 15 years of 384Kbps seems like a ridiculous jump in bandwidth to deliver. Think about a couple of phenomenon that we are experiencing right now. Firstly, bandwidth IS becoming cheaper and cheaper. I read 2 days ago that in Hong Kong, Gigabit will be available to the home for the equivalent of $220 per month. They already offer 10 and 100Mbps to the home for very low cost. Metro Ethernet is the delivery method. Whilst I acknowledge that not every city is going to be able to roll out metro ethernet to the home, it illustrates a trend. I now get 1Mpbs at home where it was 150k 12 months ago. Couple this with another interesting fact. A large UK firm has several rooms kitted out with systems from Teliris. These rooms cost £500k each approximately, require dedicated ATM connections, run at 8Mbps and are statistically less reliable than the IP and ISDN based systems. However, utilisation of these H.320 and H.323 rooms has actually waned as people want to use the high end kit. They actually say that the standards based stuff is less reliable! Their perceptions were changed by the high end stuff. This is again an extreme example but it illustrates that there is room for this stuff in an organisation.

I think that the bandwidth argument will remain valid while people still percieve that 384k is the most you should need. The move to IP should remove this glass ceiling.

Trapezhoid, you make some excellent points, technically accurate and well put.

You ask what does home have to do with it? 'TV quality' has been the measure of visual quality for videoconferencing since I started this game in 1997. I suspect it's been the benchmark of the uninitiated since well before then. Why should a user have to accept the benchmark of quality defined by our industry, when at home they see far better quality.

You ask why people haven't been using bitrates higher than 512 before now, the graph on the wainhouse report answers that. What would be the point? Even if that graph is technically inaccurate, it is pretty much bang on. It's required a series of events to pass for this graph to become untrue. The H.264 protocol, new silicon, improvements in display technology, bandwidth commoditisation, all these factors I beleive are going to enable LifeSize to make an impact.

Again fair point about the HD resolutions, but look at it from LSC's point of view, what would be the point of supporting 1080i when as you say, the displays don't support it natively. What LSC are doing is setting correct expectations, the resolution they are using is more than good enough for this application, whether you want to argue about its credentials to be called HD or not.

I dropped some hints regarding Codian in my previous post, allow me to expand, people will be opening the LSC boxes soon enough anyway. The TI chip array used by Codian is capable of delivering H.264 at 2Mbps, for up to 40 sites. Not only this but they can deliver individual layouts for each of those 40 sites. I could go on, I won't for readability. The point is that there is some serious computational power in there, several orders of magnitude more power than it's nearest rival. Wouldn't it make sense to use this? It's not the only technology in there, but it's in there.

The bus between the LSC camera and codec is capable of delivering 400Mbps. The camera has been developed by the same guy who created the viewstation, amongst other things, I think it's fair to say that he knows what's required of a camera for a videoconferencing application. I've seen the camera in development, I can confirm it delivers the goods. You are also correct, it is the most interesting development, the great leveller of VC technology has been the generic camera used on the codecs. As a friend of mine once (crudely) put, 'you can't polish a turd'!

As for the math, I respect your abilities as a engineer, but it DOES compute. I know what's inside this box. LSC haven't built their own silicon, but someone else has.

To answer your first point, Image quality does not just equal Resolution plus Frame Rate. What makes you think that Malloy et al are going to come to market with a dud product that only sends 1/4 screen updates? Be serious, these guys have got a lot riding on this, they know the expectations of both the industry and end users. Do you honestly expect that people would accept a product that performs within the parameters you describe? There are a lot of other innovations going on here that haven't been documented, for example some excellent work on colour depth, and colour 'bleed' as seen frequently in VC calls.

You said 'if it's too good to be true, it probably is'. If anything, this is the statement that fired me up. Beleive it or not, I'm a pragmatist, however once in a while something truly exceptional occurs. It annoys me greatly when people are so quick to put the boot in, especially in a situation like ours where the Tandberg / Polycom duopoly is stagnating our market.

Something that hasn't been discussed much yet is the audio side of things, particularly the phone. This is a serious peice of kit, the phone itself has the same chip that powers the VSX7000. It's not only innovative with its beam forming technology for locating a person speaking, it's also capable of far greater things. More to come on that, I wouldn't like to ruin any future surprises. What do people make of this?

To finish, I'd like to say that Polycom's announcement has only served to validate what LSC are doing. Whilst this announcement is typically knee-jerk, I mean they had to say something, they really have played right into LSC's hands here. The real question is, if and when the big players do come to market with a HD product, will the ensure it works with LSC? If past records are anything to go by, they won't, it will be a 'better' implementation and the cycle of stagnation will begin all over again. I really hope this doesn't happen.

One things for certain, it's given us all something decent to talk about again, and for that alone I'm glad, I was getting bored of firewall traversal :)

[Glen Sykes]

1. The bandwidth required to deliver the additional resolution is way beyond what is in common use today, meaning you will most likely not be able to interop these features for a long time even if other vendors actually released software upgrades for HD tomorrow.


Dead right, 1280x720 will only be available with calls between LSC endpoints right now. However, to ensure that customers don't build technology islands, it will talk to other vendors equipment. Something I'm looking forward to seeing is a call between LSC and say Polycom compared with Polycom and Polycom. With the camera LSC are using, i'll hazard a guess that the quality seen from the LSC endpoint might actually be better!


2. Standard DSPs(meaning the current state of the art system-on-a-chip" designs geared towards multimedia, which is what everybody uses) level the playing field performance-wise and every company that uses them will have a similar level of performance per DSP.


I disagree with this. Tandberg using Nexperia with the MXP have demonstrated a far greater compute capability than VSX with the Equator chip. Codian have taken it to the next level again with the Latest TI chip. It's the camera that's been the great leveller thus far.


These three points lead to his conclusion that "If it sounds too good to be true - it is". The sad fact is that if you are standards based there is very little room for high-flying super engineers to do clever things. Because of this LifeSize has done the only thing they were able to do to make waves - add a feature that looks impressive, not because it is an amazing technological breakthrough, but because it ignores real life limitations, in this case practical bandwidth restrictions.
[\UOTE]

Agreed, add to this supporting your existing user base also places further constraints on what you can do with technology, (or at least how often you can refresh) as the initial resistance to MXP will testify.

[QUOTE=Yukikaze]
Your comments about HD in the television world also miss their mark entirely. In addition to comparing apples to oranges(TV sets only decode - Videoconferencing has to encode as well, TV is broadcast and can thus afford the higher bandwith etc) you also seem to be unaware that Nexperia actually is marketed as being able to handle HD. Finding chips capable of encoding and decoding HD in real time is not a problem, finding chips capable of improving the compression ratio over existing H.264 implementations by a large amount while still being standards based is. Even if you lift the real-time restriction.

I wasn't comparing apples with oranges, it was a generalisation to illustrate the changes that are happening, and that will affect peoples perception of what video quality is. Also, it was an illustration of technology in general, and the fact that as H.264 is not just the confine of videoconferencing, as was H.261 and H.263, it is a share algorithm with broadcast media, meaning greater investment in silicon from other, non-VC players. As a point of interest, LSC do use a single nexperia chip in their codec, but it's not used for video encoding.

Finally, it's not really the chip itself that's going to improve the compression ratio, it's the algorithm. All the chip is going to do is provide the horsepower to run that algorithm within defined parameters. In laymans terms, if H.264 is the constant, then if you increase resolution, frame rate, or any other factor affecting general video quality, then the bandwidth must increase also. The only way to reduce the bandwidth is for a more efficient algorithm than H.264. This in turn requires more processing power, and so the cycle begins again.

This is why I think it's unfair to criticise LSC for using 1Mbps to deliver HD, it's not as if they had much of a choice, they are still governed by the laws of physics and the constraints of H.264.

[trapehzoid]

With respect to bandwidth, at face value 1Mbps or 2Mpbs, when compared to what the industry has used as its standard for the last 15 years of 384Kbps seems like a ridiculous jump in bandwidth to deliver.

The reason the 384 number has been so stagnant is because for the same speed, people have been getting ever increasing quality gains w/o changing their network. It has also been the most 'interoperable' speed... high chance between two random people, that is the highest speed you are going to have in common. The number isn't magic really.. it just has legacy inertia.

Think about a couple of phenomenon that we are experiencing right now. Firstly, bandwidth IS becoming cheaper and cheaper. I read 2 days ago that in Hong Kong, Gigabit will be available to the home for the equivalent of $220 per month. They already offer 10 and 100Mbps to the home for very low cost. Metro Ethernet is the delivery method.

verizon offers fiber to the house in many areas.. again doesn't mean you are going to get those speeds across the country.. or that your target reciever has the same privledge. Again.. same reason why 384 has stayed the 'norm'.. its a reasonable expectation of what you can get end-to-end. Thats why I contend that these high speeds are still only relevant in private networks.

You ask what does home have to do with it? 'TV quality' has been the measure of visual quality for videoconferencing since I started this game in 1997. I suspect it's been the benchmark of the uninitiated since well before then. Why should a user have to accept the benchmark of quality defined by our industry, when at home they see far better quality.

Well considering the entire industry is based upon lossy compression.. which nearly by definition means 'compromise', I don't agree with TV quality being a benchmark. I think 'visual presence' is better. Do people feel like the person is in the room? Do they get fatigued using the system? Do they feel like the person at the remote end was fully engaged?

Again.. I find this fancination with HD so quickly a mockery. How many people were just preaching how desktop is the eventual way.. desktop and commodity pricing. With one annoucement from the 'chosen ones' all of a sudden everyone flips? Lets merge the two concepts...
What good is HD at the desktop? If you look even at a CIF image on a 8" or so screen it looks incredible. You don't need 10x the resolution on a screen that small. Nor are you going to get that kind of bw in those kinds of numbers.. when talking about deploying hundreds or thousands of units within an organization. So how do these annoucements about HD further push the industry towards those previous goals?

You ask why people haven't been using bitrates higher than 512 before now, the graph on the wainhouse report answers that. What would be the point?

put down the koolaid! What's the point? Have you forgotten everything you've seen since before the LS annoucement? Make a call at 384kbs and then make a call at 2+ megabits. Remember.. lord.. its COMPRESSION that is still hindering quality.. not just resolution. Speeds make HUGE differences in quality today.. even at the same resolution. So to argue that people have not used higher then 384kbs based on the fact that resolution doesn't increase is flat out absurd. The graph in the wainhouse article is a LS chart.. nothing more.. its intended to show a products features, not facts about technology.

Even if that graph is technically inaccurate, it is pretty much bang on. It's required a series of events to pass for this graph to become untrue.

What's true about it? people offer higher then CIF resolution today.. just not at 30fps. I bet people could do it, but probably haven't done it because of the bandwidth required. Shame on them I guess for not doing it for just 'doing it' I guess. in hindsight doing it probably would have taken the wind out of some of LS's sail. But doing higher resolutions requires more frame memory.. probably not something all vendors can do. But many systems can do XGA now, so probably wouldn't be a stretch to crank up the frame rate. Just before it was probably just 'why? its not where the target users are'

What LSC are doing is setting correct expectations, the resolution they are using is more than good enough for this application, whether you want to argue about its credentials to be called HD or not.

720p is certainly HD.. but its certainly being exploited in this case. Note how they don't refer to it as 720p? but rather the higher looking 1280x720 resolution? Because it sounds 'better'.. its psychological marketing.

To answer your first point, Image quality does not just equal Resolution plus Frame Rate. What makes you think that Malloy et al are going to come to market with a dud product that only sends 1/4 screen updates? Be serious, these guys have got a lot riding on this, they know the expectations of both the industry and end users.

Because you can't ignore the math of it. There are probably clever tricks to be able to display and update that much resolution without actually sending as much as if it were a typical CIF or 4CIF image. How do you think polycom's and tandberg's packet loss stuff works? Its through creative methods to make the image more resilant to loss through segmenting up the picture so when some is lost, it can be easier to mask. You could use similar concepts to break down how much of the image you need to update at once to deliver this high display resolution without necessarily encoding such a high resolution all the time.

How was the end to end latency in your demo? Did you notice at all? How about when you start trying to use it where there will be real world network latency involved? Oh wait.. that will not be an issue as who makes 5mbit calls transatlantic :D

Do you honestly expect that people would accept a product that performs within the parameters you describe?

People do strange things. I'm not here to argue the intelligence level of the world's consumers :)

It's not only innovative with its beam forming technology for locating a person speaking, it's also capable of far greater things. More to come on that, I wouldn't like to ruin any future surprises. What do people make of this?

I predict plcm to silently challenge them greatly on the patent side of audio locating once they get a chance to look at a unit. PLCM has a strangle hold on the conference phone market.. I would not expect them to sit idly by why another player tries to interject.. especially ex-employees.


To finish, I'd like to say that Polycom's announcement has only served to validate what LSC are doing. Whilst this announcement is typically knee-jerk, I mean they had to say something, they really have played right into LSC's hands here.

Validate? if anything it validates things about polycom and their business beliefs.. not LS at all. they have a history of 'me too' annoucements.. I didn't even bother reading the polycom HD press release.. I knew what it would say.

The real question is, if and when the big players do come to market with a HD product, will the ensure it works with LSC? If past records are anything to go by, they won't, it will be a 'better' implementation and the cycle of stagnation will begin all over again. I really hope this doesn't happen.

Well.. I would imagine that if the other players find the LS methods inferior.. why would they cripple themselves to match LS? I see NearTV and siren14 as good examples of that. siren14 has a place, so maybe that's not the best example.. but it certainly isn't the cats meow and why other people didn't jump to support it... even after polycom offered licensing models for it, etc.

The good news in this is if they are smart, they are using H264 and standardized annexes.. if that's the case compatibility will mainly be an interop and supported resolution thing... so other vendors should hopefully not have to bend over backwards to support LS's methods.

Wrapping up.. why people have not gone this way before for higher resolutions I contend has been the lack of good quality cameras and cheap displays. The market has been moving towards commodity, not high end. Even as sources such as document cameras and others started trying to offer the higher resolution progressive images they still sucked for motion video. I think its that area that has kept VTC makers in check and happy with 4CIF or less for face cameras... which is why I say the camera if it plays out is the most interesting part of LC's annoucement.

The cameras plus the cost of high resolution displays were reasons to say happy at CIF and 4CIF. But the falling prices of projectors and plasmas have really made high resolution display more available.. but high resolution input has still elluded the market.

[Gary Miyakawa]

Actually, I, personally, would be pretty happy with 30fps (60 fields) at NTSC/4CIF at or near 768/1mb.. This would allow us to use the current technology cameras and monitors with near TV resolution.

I know the Sony does it.. I've tested it on the Polycom VSX7K series and was able to get 15fps @ 4CIF @ 1mb...

Seems like not alienating the current technology might be worth while and cost effective..

Just my thoughts..

Gary Miyakawa

[Glen Sykes]

I'll try to keep this answer shorter than the last :classic:

I'd agree the reason why resolution has not been an issue until now has definitely been because of limitations in camera and screen technology, and pricing. Now that HD is rolling out to homes, the benchmark for peoples perception for what consititutes high quality video is about to change. For that reason, I think that what Lifesize is doing is valid.

The other limiting factor has been bandwidth, and one other major reason left out by Trapezhoid for the longevity of 384 has been ISDN. The self imposed limitation of 384 (which makes absolute sense) on ISDN generated the need for better compression, better overall performance over a given bandwidth. I do beleive though now that we are at a watershed, where IP bandwidth is becoming more and more a commodity, not only in domestic but also in business, and that the 384k limit is now no more than a glass ceiling.

As a matter of course, we will see the introduction of more efficient encoding algorithms, coupled with hardware capable of using them resulting in bandwidth savings, and this can only be a good thing as video communications tend toward mass usage.

To answer something specifically, you're correct about the latency, there was a visible lag in the direct comparison between 2 x LSC rooms and 2 x VSX7000 at 1Mbps. This latency will be improved upon, bearing in mind that there are several months before these things ship. The picture quality was outstanding, and given the choice it was a no brainer which system was the better to use, and this demo was still waiting on some significant camera improvements that were yet to be implemented.

I think you're correct about one thing though, 5Mbps point to point transatlantic calls are not going to happen, it's just gratuitous, but I don't see 1 or 2Mbps as being a problem providing the network is up to it.

On the audio conference phone side of things, I'm not sure how many polycom patented techniques LSC are using, however the beam forming technology is completely new, LSC already own several patents on the technology in the phone. One things for sure, if they are using Polycom patents, they

I'm still intrigued by Trapezhoid's claim that they cannot be sending full screen updates of 720p at 1mbps. What are you basing this on, bearing in mind that I've already said that there is some brand new silicon in this box.

I agree with Trapezhoid on the desktop situation however, HD will have a very limited affect on this, however I think it's unfair to make this comparison, given LifeSize's story is really about meeting room quality, 'Life Size' images on large screens in high resolutions, where a viewer sat close to the screen doesn't have to look at a picture of someone like they were made out of Lego.

Desktop video (indeed desktop communication per se) is more about spontaneous, bursty conversations, rather than structured and time consuming conversations, and in this case, higher resolutions are a nice to have, much further down the priority list.

[Sean Lessman]

Maybe you guys see a different market than I do or have different customers than I see. We do not receive tens of calls a day on why the video doesn't look 'better'. We receive tens of calls a day on why the customer cannot connect (i.e. firewalls).

I would love to see video get better, but I can honestly say I think the latest products on the market have excellent video especially at 1Mbps and above.

I think solving the connection problems (FWT/NAT, calling plan), reliability (inbuilt transcoding, downspeeding, ratematching), and availablity of features at all time throughout the endpoints and infrastructure are more important than incremental improvements in video. I cannot remember ever hearing a customer say "had to use the phone, because the video was unacceptible in quality (unless there is packet loss etc)". I have heard customers say "had to use the phone because I could not get the video system to connect". But thats why I work at TANDBERG, I believe in the direction we are heading.

Better video is nice, but it doesn't do much if you cannot get the call up and keep it up.

While most seem to be celebrating the announcement, it disappointed me. If we continue to be amazed by a something like incremental improvements in video and do not see the value of the bigger solution -- we have a long way to go as an industry. Just my $.02.

Congrats to LifeSize on a big entry to the industry and I look forward to their contributions.

Also, bonus question, anyone have any idea what bandwidth DirectTV uses to transmit HD 'quality' to its customers? :-)

Sean

[trapehzoid]

The other limiting factor has been bandwidth, and one other major reason left out by Trapezhoid for the longevity of 384 has been ISDN.

I didn't leave it out.. I thought it was self evident that I was talking about ISDN as the 384 number was around h323 even existed. When IP came around.. we have the stupid arguments about 'you need 2x the IP bandwidth to get the same quality as ISDN' which was rubbish. People confusing full duplex vs half duplex. Then people started using 768kbs commonly on IP, but you see again.. as people start moving towards talking BETWEEN companies.. and not just internally the 384kbs number is still kicking.

I do beleive though now that we are at a watershed, where IP bandwidth is becoming more and more a commodity, not only in domestic but also in business, and that the 384k limit is now no more than a glass ceiling.

Ok, here's a simple litmus test for yourself. How much has the internet connection bandwidth at your company increased over the last 2 years? What instigated those changes? Did anyone add more bandwidth simply because its cheaper now?

As a matter of course, we will see the introduction of more efficient encoding algorithms, coupled with hardware capable of using them resulting in bandwidth savings, and this can only be a good thing as video communications tend toward mass usage.

Ok, how has anything you just said above validated your arguement that requiring 1+mbit is valid for the market for a moving forward position? What you just said is exactly why bandwidth requirements have been going DOWN and being driven down by demand.

I think you're correct about one thing though, 5Mbps point to point transatlantic calls are not going to happen, it's just gratuitous, but I don't see 1 or 2Mbps as being a problem providing the network is up to it.

Ok.. I'll return to my previous test. Lets look at your own company where you work. How much bandwidth do you have generically to the Internet? I can tell you my company uses VTC.. and I can not sustain more then 1 megabit outside my office.

On the audio conference phone side of things, I'm not sure how many polycom patented techniques LSC are using, however the beam forming technology is completely new, LSC already own several patents on the technology in the phone.

That's pretty quick. Ever think about asking for the patent #s? Awarding patents isn't the fastest process on the planet.

I agree with Trapezhoid on the desktop situation however, HD will have a very limited affect on this, however I think it's unfair to make this comparison, given LifeSize's story is really about meeting room quality, 'Life Size' images on large screens in high resolutions, where a viewer sat close to the screen doesn't have to look at a picture of someone like they were made out of Lego.

You are so inconsistant. One.. Large Screens are not INTENDED for people to sit close to. Two.. so even in your defense of the concept.. you re-enforce exactly what I am saying.. this is at best a niche market which does nothing to further advance where the mass industry is going. Large, high-end, high bandwidth rooms are not where the market is heading. yes there are applications for this, but certainly not dominate, and certainly not the leading edge of the market.

Desktop video (indeed desktop communication per se) is more about spontaneous, bursty conversations, rather than structured and time consuming conversations, and in this case, higher resolutions are a nice to have, much further down the priority list.

See this is where you don't understand the demand for desktop I think. Its not about structured and time consuming or not. Its about access, ease, cost, and instant access. Its the same concept of having the 'town' telephone vs everyone having their own.. all the way to each person having a mobile. The quality or content of the conversation didn't necessarily change.. its about access, cost, and ability to instantly reach anyone. Its those things that drive the 'quest' for desktop.. because people believe something on the desktop will solve those needs. Its not as much that desktop is scheduled or time consuming.. its that 'personal' vs 'communal' will address the ability to make video as mass deployed as telephones are.

[trapehzoid]

Maybe you guys see a different market than I do or have different customers than I see. We do not receive tens of calls a day on why the video doesn't look 'better'. We receive tens of calls a day on why the customer cannot connect (i.e. firewalls).

I like that comparison.. that makes clear what are obstacles to adopting video. I think people in general are amazed by talking to people remotely live period.. enough that the quality thing is overlooked.. only in some cases where the image itself is more important then the fact you are meeting (like medial, etc). Quality usually comes after the decision to implement video.. not before. I don't recall talking to anyone who said 'we didn't implement VTC because the quality wasn't good enough'. It affects purchasing decisions, but doesn't make/break VTC as a whole.


I would love to see video get better, but I can honestly say I think the latest products on the market have excellent video especially at 1Mbps and above.

I agree.. I can easily watch taped content at length when using high bandwidths.. and typically forget its over VTC after 10-15 minutes. Yes, I can still see artifacts.. but their existance fades from importance as I'm drawn into what I'm actually watching.

[Sean Lessman]

I see Direct Visual is now a LifeSize dealer. Explains the passion! :)

[Glen Sykes]

;)

We're passionate about everything we sell, including Tandberg :D