Good job, Derek. You just made a lot of AOL dialup users happy. :)
Testing IP Video - Super Low Bandwidth
**** *****, ***** *** ****** *********, especially ** ******** ************, ***** ********* limited ********* ** *********.
******* ****, ***** **** ** ** able ** ******* ***** ****.
** * ***** ***** ** ***** is *** ****, **** *** *** do **** *** **** **** ** Kb/s ** ****?
* ******* ** ******* ***** '*****' compression ************ *** ***** *************** **** *** ********* *********** *** expensive.
**** *** *** *** ******** **** this ****:
- *.*** ** ******* *** ** **** resolution (*******), * *** *** **** compression *** ** ******** **** * *** ****** **********, **** ******** ** *** ** 12 **/* ** *** *****.
- ******** ** ** ***** ****** ** least *** **/* *** **** *** with * ******, ** ****** ***** at * *** *** ****** ***********.
- ********** *********** **** *** ** *** **** marginally ******* ******* *******, ***** ********** bandwidth *********** ** ~**-~**% *** ****, and ~**-**% *** ****.
***************
***** *********** ** * *** ****** ********** (*********** ** ** **/* ****** max) *** ** **** ** ****** live ***** ***** ******** ******* *** CODECs, ** *** **** ** ********** and ***** ****. *********** ** ***, QVGA, ** ***** *** ********** ** ** *** ***** ****** ** used ** ****** ***** *** **** a *** ********* **********. ************, ***** motion ********* ******* ** ** **** as ***% ** *** *****, ******** should ** ******* ** ******* *** this ** ******* *** ****.
***********
***** ** **** *****, ** ******** ********* between **** *** *************** ***** **, *, and * ****** *** ******, *** compression ****** ** *** *** ***. We *** *** ****** ** ** view * ***** ****, **** *** ******* walking ******* *** *****. *** ***** of **** *** ******* ** *********** below:
**** *** ******* ******** ** *** field ** ****, ** ****** *** clear details **** *** ******* ** *** QVGA image, *** ** *** *** * general **** ** ******** *** ********. The **** *** **** ** **** scene **** *** ****** *** **** ~11% ** ****.
********** *****
*** ****** *********** *** ******* ** ** **/* (using *.** ******). *** ******* ******* frame rate/compression combination ** *** ***** **** *** not ****** **** ********* *** * FPS ** ***, ********* ***** ** Kb/s. ***** ** *** ***** ** our ******* ****** ******* *** *****.
*** ****** ************* **** * *** rate ** ** **/* *** ******** using **** **********, *.***, **** *********** ** 3 ***.
** * ******-****** ******* *** ******* *** lowest *** ***** ********, ** ******** 32 **/*, ***** *** *** ****** set ** **** **********, *.***, **** *********** at ** ***.
Bandwidth ************
**** *** *** ******** ********* ************ taken **** *******. ****** *** ***** QVGA resolution **** ******, ***** **** ** the **** ***** ***. ** *** clearly *** * ~** ********** ******* 720p ********** *** **** ** **** ********** many ** *** ********* ****** ******.
***********
** *** **** **** ***** *** Axis ***** ***** ******** ******* *.**.**.
*********** ******* *.*.*.***** *** ****.
So glad to see such good lighting conditions and high contrast between foreground / background. I bet that made the compressor super happy. :) :) :)
TeleEye do it much better, using narrow bandwidth , error correction,...they use vector quantisation not transform or wavelet based codec...
Technical info about their compression here.
I try SMACM, HD SMACM, H264, MPEG and SMAC M is the best for streaming using narrow bandwidth.
When is that technical paper from?
The codecs mentioned (like MPEG-1 and MPEG-2) point to it being 10 years old.
How do you know it would be much better? Did you test it head to head against a modern H.264 implementation? Did you validate that TeleEye could deliver lower bandwidth but with the same quality?
I am in CCTv since 1989...
Super low bandwidth applications are a niche, so there is not a lot of interest overall.
TeleEye is not a well known company so there is not a lot of interest in them.
Plus its proprietary and, as you acknowledge, expensive.
Finally, lots of people claim to have super low bandwidth codecs but it turns out they are just lowering quality level / increasing compression.
All of this makes me skeptical and not interested in testing this.
Hi Sr?an,
Yes I try H264, compared with SMAC-M
Care to post your detailed findings for us somewhere on the web? I don't mean TeleEye's findings. Their findings here appears to compare HD SMAC-M to MPEG-4 (Not MPEG-4 Part 10 or AVC) rather than H.264.
I am always open to new information so for the sake of learning, please provide more details.
Yes SMAC M is old but is better than H264
Based on compression alone? I will refer you to my thoughts on this following your statement below.
Many suppliers offers dual or triple codecs but do not tell the whole story: user can chose what codec but not in the same time. In SMAC M you can use 4 different bitrates in te same time
- I know of a Tier 2 to Tier 1 product that offers 4 streams all using different codecs and frame rates. Some of their products might even be considered Tier 3 and even these inexpensive cameras offer 3 simultaneous streams.
- Secondly, unless all VMSes supported HD SMAC-M your point is moot to the vast majority out there that use the well-known VMSes for very good reason I.E recording, user, analytic, open & other features.
- H.264 is a direct successor to H.263 and it is leading to H.265 developed by the "ISO/IEC Moving Picture Experts Group (MPEG) and ITU-T Video Coding Experts Group (VCEG) as ISO/IEC 23008-2 MPEG-H Part 2 and ITU-T H.265". Theforefore, we can expect the industry to move in this direction.
- Given the above, development will be quicker and much more widely supported and adopted than any proprietary encoding method unless such method is supported by say Google - such as WebM (which at this point still trails H.264 as we speak, but WebM is also developed to be used on the web for HTML 5 primarily). Even this has not taken off completely just yet.
With that said, there are some benefits to proprietary codecs and I do not argue that. To give you an example: Geutebrück uses MPEG4CCTV and H264CCTV which are different iterations of the technologies they are based on. Still, the VMS supports normal MPEG-4 and H.264.
The downside is their proprietary technology is only natively available on their cameras unless a transcoder is used.
The upside is that changes in pictures are not referenced to the I-Frame but rather the previous B-Frame. This is not a major advantage, but it helps in the event of a dropped connection or an unstable connection. Normal encoding will reference as far back as to the last I-Frame with all subsequent changed scenery lost.
Still, the technology can only be used on Geutebrück thus making it less relevant.*
I'd venture to say future implementations of encoding mechanisms will resolve most issues exposed by proprietary encoding methods anyway.
---
* - Although the VMS allows pulling RTSP from the Geutebrück platform so there is a minor exception. The feature is mostly used for AD-HOC requirements.
All this paper says is that SMACM is an encoder system that outputs multiple video streams and different bit rates, and that (somewhere in there--at all bit rates?) SMACM uses a vector quantization based codec.
Any codec can be configured to output video at lower bit rates. The question is which one might provide the best compression.
Vector quantization enjoyed some popularity 10-20 years ago becuase it made the decoder's job much easier at a time when CPU resources for decode were precious. But since clients have become more powerful the VQ algorithms have fallen off in favor of more efficient encoding in MPEG et. al.
Also, being a proprietary system is going to limit the audience of this kind of technology--especially among the IPVM crowd.
At a previous job I had a customer with the same requirement but it was for a PTZ camera on a remote highway and they needed an encoder / decoder pair to transmit the video over the 56k line. I could compress the video but the latency was in the range of 5-10 seconds and made it all but impossible to move the PTZ.
Did you measure latency in this test? Does the user need live video or was it good enough?
I posted without my name to protect the innocent :-)
Latency is a good point. I don't think latency is likely an issue in this particular testing. But in an end-to-end system buffering can sometimes help achieve stable streaming at lower bandwidths by accommodating variability in the bit rate of the communications channel. Since streaming has a 'real time' requirement (that bits arrive at the decoder at a rate at least as fast as the video is encoded at) the illusion of a greater encoded bit rate can be achieved if buffering is used to accumulate video when the channel has a higher bit rate and drain slightly when the bit rate is low. But buffering of course introduces latency that is less desireable in the case of something like PTZ.
We just done some tests with a well known VMS that has dual streaming so we could setup 1.) high quality for recording and 2.) lower quality for remote vieving. These streams can also be allocated bandwidth, Frame rate and Resolution and compression limits in the VMS on the selected CODEC. You can also click the playback button and digitally zoom into the HQ image.
Additional this software has the function to down sample frame rate and quality to client, so I would think this is a 3rd quality stream from the system. The system also has a 4th stream from a mobile server on the recorder for mobile viewing.
I believe similar to Avigilon, this is Teleeye's way of marketing something unique, that when pulled apart is not really that unique. In actual fact there are many products that have similar features and could most probably achieve the same desired end result for the customer without having to buy preprietary cameras.