PDA

View Full Version : Wainhouse on Polycom's H.264 High Profile same user experience quality at BW reductio


Rodolfo_Dávila
11-19-2010, 09:29 AM
Hi people,

I was reading the Wainhouse Research whitepaper about Polycom's claim about H.264 High Profile BW reduction with the same user experience.

Wainhouse's conclusion:

"Polycom’s implementation of H.264 High Profile allows for bandwidth / call speed reductions of up to ~ 33% with little or no impact on the user experience– even in relatively high motion situations"

"Polycom’s implementation of H.264 High Profile allows for bandwidth / call speed reductions above 33% (e.g. 50 – 60%), the impact on the user experience ranged from very limited (in low motion / talking head situations) to notable but tolerable (in high motion situations). Whether or not the user experience compromise associated with this drastic bandwidth reduction represents a problem depends entirely upon the expectations of the video meeting participants."

source: http://www.wrplatinum.com/Downloads/11977.aspx


It's hard to believe, in my opinion, that a BW reduction using HP will not affect the quality compared with BL. What's going on with latency testing? Wainhouse based that issue in a subjetive assessment.

What's your opinion on this Wainhouse study?

Regards

Sean Lessman
11-19-2010, 09:46 AM
Keep in mind this study was comparing Polycom to Polycom, so the conclusion is basically Polycom HiP at half the bandwidth looks the same as Polycom BP. Not really sure what that tells you.

Rodolfo_Dávila
11-19-2010, 10:03 AM
Yes, a Polycom's H.264 HiP close solution.

I don't see a real advantage in terms of interoperatibility in this heteogeneus VTC world. In my company we have more than 50 endpoints( VSX, MXP, lifesize, etc) then in my case I don't see a great advantage if I can interoperate.

If Polycom wants to interoperate with another manufacturer(supposing that it supports HiP too), there must be a signaling compatibility and different flavors for each manufacturer.

Rodolfo_Dávila
11-19-2010, 10:07 AM
...if I CAN'T interoperate

Sorry

Sean Lessman
11-19-2010, 10:41 AM
What I don't get is how they can say

Independent tests conducted by industry analyst firm Wainhouse Research (http://www.wainhouse.com/) validate that Polycom (http://www.polycom.com/index.html), Inc. [Nasdaq: PLCM], a global leader in unified communications, more than delivers on claims that Polycom telepresence solutions deliver high-definition (HD) telepresence calls at up to half the bandwidth of Cisco and other major competitors – and with no appreciable sacrifice in call quality.

in their PR, when they only tested against themselves.

PR: http://www.polycom.com/company/news_room/press_releases/2010/20101118.html

Sean

Rodolfo_Dávila
11-19-2010, 10:56 AM
Polycom's marketing desperate move, trying to confuse the client with that kind of stuff

Tigger09
11-19-2010, 04:47 PM
How independent can Wainhouse be if Polycom is paying them to cunduct the test?

where is Cisco\Tandberg in all of this? How come we haven't seen them show proof and\or numbers that HiP isn't what PLCM says it is?

jwinterb
11-19-2010, 08:58 PM
Hi there

Google

"Context-Based Adaptive Binary Arithmetic Coding in the H.264/AVC Video Compression Standard"

All the research has already been done on this. It's a bit of a read but the results show more like a 9-14% bitrate increase a little lower that 50% or 33%

Its also important to note that this is only possible in a Polycom to Polycom call.

I don't really know what to think about this announcement. I'm all for innovation and advancing the video industry but what I don't like is adopting a standard that doesn't interop with the rest of the world and then claiming that the rest of the video industry is behind the curve because they haven't adopted the same standard. Are we going to start throwing all non AVC supported systems into the "Legacy" pile now?

jpd
11-20-2010, 02:23 AM
I don't really know what to think about this announcement. I'm all for innovation and advancing the video industry but what I don't like is adopting a standard that doesn't interop with the rest of the world and then claiming that the rest of the video industry is behind the curve because they haven't adopted the same standard.

Mmm, this latest blurb from HP (http://www.telepresenceoptions.com/2010/11/hp_expands_video_offerings_wit/); grabbed my attention, particularly;
quote
...video network infrastructure components capitalizes on the SVC video standard..unquote

Sean Lessman
11-21-2010, 04:46 PM
...where is Cisco\Tandberg in all of this? How come we haven't seen them show proof and\or numbers that HiP isn't what PLCM says it is?

Don't have to...there are plenty of 'real' independent papers out there by the IEEE etc. that would suggest PLCM has taken a number of liberties on their claims.

Sean

Sean Lessman
11-22-2010, 09:50 AM
We set up some tests today to reproduce what Wainhouse tested. It is clear that the HiP does improve Polycom video, but for fun we set up the same products in a 4-way side by side (1. Live video 2. C-series at 512kbps 3. HDX at 512kbps 4. HDX at 512kbps HiP). We would believe the C-series at the same bandwidth does as well or outperforms HiP from the HDX. We tried the same thing at 1Mbps and 4Mbps and feel very comfortable that the Cisco (TANDBERG) endpoints running baseline profile perform as good or better than HDX HiP at the same speed. We also compared their claims of half the bandwidth and it wasn't even close.

I would definitely challenge the claim that PLCM outperforms Cisco at half the bandwidth when they cannot outperform our products at the same bandwidth (our baseline profile vs their High Profile). Using HiP however does greatly improve PLCMs video over their own baseline profile codecs, so it is a net improvement for their video when compared to only PLCM.

If you are considering a hardware upgrade to get HiP from PLCM, maybe you should consider a manufacturer upgrade at the same time.

Sean

Rodolfo_Dávila
11-22-2010, 04:13 PM
Using HiP however does greatly improve PLCMs video over their own baseline profile codecs, so it is a net improvement for their video when compared to only PLCM.

Great information Sean, thank you very much for your valuable mythbuster's tests

That's exactly the point here, Polycom improve over itself(not over all the market) and they paint all that stuff like if it was "the VTC discovery nobel price".

The claim "HD at half BW" is good for Polycom's solution, but when is compared with what Tandberg's baseline profile can achieve, it shows that Polycom is exagerating.

Sean Lessman
11-23-2010, 11:12 AM
I also found that HDX with Baseline Profile had a higher latency than the TANDBERG C-series...and HDX High Profile had an even higher latency over HDX with Baseline Profile. Looks like High Profile in the HDX has significantly increased the latency. Guess Wainhouse missed that one since the explicitly stated they didn't test latency (but isn't that user experience?).

Sean

Gary Miyakawa
11-23-2010, 01:18 PM
I also found that HDX with Baseline Profile had a higher latency than the TANDBERG C-series...and HDX High Profile had an even higher latency over HDX with Baseline Profile. Looks like High Profile in the HDX has significantly increased the latency. Guess Wainhouse missed that one since the explicitly stated they didn't test latency (but isn't that user experience?).

Sean

Sean,

Those are interesting comments about latency since I did analytical tests with the code (admittedly, it was 2.5.x back on 4/7/2010) and I posted here that I found there was virtually NO difference (less than 5% either way) across all bandwidths and more that 4,000 samples...

I'd love for Tandberg(Cisco) to provide me with two C series products and I'd be happy to run the same analytical tests (I still have all the materials) and post the latency results.

It would be very interesting.

Gary Miyakawa

Tigger09
11-23-2010, 03:42 PM
So at what point does a PLCM HiP call to an RMX start to look like a satellite VTC call?

:)

Sean Lessman
11-23-2010, 05:53 PM
Those are interesting comments...

Gary, it was visibly noticeable...didn't require a sophisticated setup like Wainhouse likes to use normally to see the latency differences between HDX baseline profile and HDX high profile. The C-series beat both in latency, it is clear as day just using your eyes. HDX high profile has higher latency than HDX baseline.

As I said before the real data point is that a C-series using plain old baseline profile outperforms a HDX with High Profile at the same speed...forget half the speed, its even worse. I have no problem with anyone doing a side by side, go for it.

Sean

Sean Lessman
11-29-2010, 11:18 AM
Check it out...

http://www.youtube.com/user/VideoIsTheNewLoud

This is a series of side by sides of a C-series endpoint using baseline profile vs a HDX 8000 using High Profile. Which experience would you prefer?

Take the challenge: http://www.youtube.com/user/VideoIsTheNewLoud#p/a/u/2/_pDNlGODJlE

Sean

Rodolfo_Dávila
11-29-2010, 03:53 PM
Great test Sean, thank very much. It's clear, the Polycom "HiP fantasy" can look great on a datasheet, powerpoint, etc.

Wow, looking at part2, the comparative is so clear, the experience of the Polycom's HiP is the worst.

I don't understand how a research company like Wainhouse can declare that kind of lie, declaring that the experience of the Polycom's HiP it's the same comparing with Polycom's BP

Rodolfo_Dávila
11-29-2010, 03:57 PM
...Or even worse, Polycom declaring that HiP is better than Cisco(Tandberg) BP. Yeah right....:tired:

Tigger09
10-19-2011, 09:33 PM
10Just buy a Edmonton ****** (http://www.asian******sedmonton.com) smaller vibrator as to Edmonton Asian ****** (http://www.asian******sedmonton.com) not cause discomfort. It can Edmonton ******s (http://www.asian******sedmonton.com) go in as far as you Edmonton Asian ******st (http://www.asian******sedmonton.com) want it to. I find immense pleasure with smaller vibrators. I used to have this huge one that did everything I found it was just too intense for me, and too big!

I'm sold, where do I sign up?