Smart Codec Guide

By: IPVM Team, Published on Feb 01, 2018

"Smart Codec" was once seemingly a marketing buzzword, but in 2020, nearly all new camera models include this tehcnology.

These marketing names vary, including 'Zipstream', 'Smart Coding', 'H.264+', 'Smart Stream II', and others, and critically, these implementations and bandwidth savings vary dramatically.

In this Guide, we explain what smart codecs attempt to do and the most common implementations, covering:

Historic static compression, I-frame interval, and FPS techniques
Smart codec basic
Dynamic compression
Dynamic I-frame interval/GOP
Dynamic FPS
Static compression regions
Intelligent DNR
VMS/NVR compatibility
Manufacturer support
IPVM test recommendations
Bandwidth risks
Smart codec outlook

To understand this, you must have a good understanding of codecs, compression and bandwidth variations. Please first review our:

'Normal' codecs, like 'regular' H.264, set one compression level, one I frame interval, and one frame rate. 'Smart' codecs change one, two, or all of those, reviewed inside.

"Normal" ***** ******

** ****** ******, ***** key ********** *** ********* set ** ***** ******:

*********** ***** (********* ****** quality ** ************)
* ***** ******** (********* called *** ** ***)
***** **** (***)

**** ** *** *** not ***** ** *****, they ***** **** ************* making ***** *** ********* on *** ******** (*.*., see** ****** ************ *********** Comparison).

Fixed *********** *******

'******' ****** *** * single *********** ***** *** the ****** *****, ********** of **** ** ***** displayed. *** *******, ** we **** ** * scene ** * *******, both *** ****** ******* as **** ** *** static *****, *****, ****, and ***** ********** ***** will ** ********** ** the **** ***** ** each *****:

Fixed *-***** ******** *******

*** **** *********** ********** between *****'* ****** (*.***/*.***) and ******* ****** **** MJPEG ** **** **** do *** **** *** 'same' ***** **** *** over *****. *******, * single **** ***** ** sent (****** ** '*-*****'), followed ** **** ***** updates ** *** ***** of *** ***** ***** have *******, ****** '*-******.' In ******** ******, *** I-frame ******** ** *****. This ***** **** *** camera **** ****** ******** an * ***** ************, most ******** * ******. In **** ********, ** the ****** ** ** FPS, ** **** **** 1 *-***** ******** ** 29 *-****** *** **** repeat.

*** ******** ** * fixed *-***** ******** ** that *** ******** ** the ***** *** ****. Sometimes * ******* *** be ***** *** ** hour. ****** *** '*****' time, ******* ** * frame **** *** ****** is ********. ******** **** I ****** ******* *** more ********* **** * frames, ***** ** *** order ** *** ****.

*** **** *** ***** of ****** **** *******, I ***** ******** ****** be ********* ** ****** artifacts *** ***** ****** (see****: *.*** * ** P ***** ******).

Smart ***** **********:

****** ******* ******, ***** codecs *********** ****** ***** parameters, ***** * *********** of ***** **********:

******* ***********:******* *********** ** ******* areas ** *** ***** instead ** *** ****** frame.
******* *-***** ********:******* *-***** ******** ********* on ****** ** *** scene.
******* *********:******* *** ********* ** motion ** *** *****.

** **** ** ***** in **** ****** *****.

Dynamic ***********

*** ***** ********* **** is ******* ***********, ***** adjusts ************ ***** *** different ***** ** *** scene ***** ** ********. For *******, ******* ** setting *********** ** '******' for *** ***** *****, as ***** ** *** hallway ***** *****, *** camera ******* *********** ** the ****** ******* ** low (****** *******), ** maximize ******* *******, *** increases *********** ** ****** background *****.

** ********* *********** ****** for ***** ** *** scene, ** ** ******** to **** ****** ******* bandwidth ******** *** ********* quality ** ***** ***** matter ** **** ************* assigning *********** ****** ** relevant ***** ** *** scene.

Dynamic *-***** ********

******, ***** ****** *** adjust *-***** ******** ***** on ******** ** ******** in * *****. ** long ** *** ***** has ****** ** ******, the ****** *** **** an *-***** ************ (*.*., every * ** ** or ** *******), *** as **** ** *********** activity ** ********, ** I-frame ** *********, **** the ******** **** ***** as **** ** ******** continues.

*** ***** ***** ***** analysis ** * ******** H.264 ****** ****** *** using ******* *-***** ******** (GOP). ***** *** ******** codec ******* ***** ** an ******** ** **, the ******* *** ******, increasing ** **** ** when ***** ** ** motion ** *** *****. Note **** **** ******* shown ***** **** ** FPS.

***** ******* *-***** ******** will *** ******* *******, quality ****** ****** *********** the ****, ***** ************* reducing *********. ** *** tests, ******* *** ** the ******* ****** ** bandwidth ******* ** ***** codecs.

Dynamic ***

****** **** *****/**** **** included ******* ***, ****** '****** ***** *********' *** ******* *****, its ************ ** ** cameras ** **** ******. Like *-***** ********, ******* FPS ******* ******* ** simply ******* ***** ****** when ***** ** ** activity ** *** *****.

***** **** *** *********** reduce ******* (********* ***** 10 **/* *** ******** in *** *****), ***** should ****** ** ******** frame **** ** **** moving ******* *** ** missed ** **** *****, such ** ******** ** highway *****, ******* ******** in ***** *****, ***. Additionally, ** **** ***** regulations *** ******** **** frame **** *********.

Not-So-Smart: ****** *********** *******

**** ******* ***** ***** to *** ***** ***** as ********** *** **********. This *** ****** ******* in **** ***** *** introduces ********** ***** ******** to ***** ******:

************* **** ***********:** ******* *** *** carefully *** ***** ** actual *** ** *** area ***** ********, ****** objects *** **** **** higher *********** "**********" *****, potentially ****** ****** *******.
******** ******** ** ********** change:** *** ***** *******, users **** ******** ****** their ****** ******* ** properly ******* *** ** the *****. ** ********** is ******** **** ***** smart ******.

Related ***** **********: ***** *********

****** ****** **** ******* with ******* *****, ****** in *** ***** ******. This ***** '*****' ****** the ****** *** ********* tricks ******** **** ******** it ** **** ********. Because *** ****** ******** it ** ********, ** takes **** ********* (***** a *** **** *********). See**** / *** *** Video ************ *****

************ ******* *** ****** intelligence ** *********** ******* real ******** (* ****** walking, * *** *******, etc.) *** ******* ***** which *** ****** ********* significantly. *** ****, *** our****** *** (******* ***** Reduction) *****.

Works **** ******** *****/****

******* ***** ****** **** within *** ******** ********* of *.***/*.***, ********** ******* is ********* *** ******** by ***** *** ****, in ******** ** **** specialized ****** ***** ***** to ****** ********* (**** as******* ************** ********' *******) *** ******* ****** integration.

**** ***** *** ********** issues **** ***** ******** or ******* *** *******, but **** *** ********** have ******** ******* ***** smart ******' ************ *** removed ***** ******.

Manufacturer ****** *******

******* *********, **** ******* a ****** ****** *** dynamic *********** ******, **** not **** ** ******* that ** ****** *** smart *-***** ********* ** dynamic ***. **** ******* only ******* *** ** the ***** (**** ******** dynamic ***********), ***** ****** support *** *****.

*** ***** ***** ******* which ************* ******* ***** smart ***** ********:

Manufacturer *********** ******

**** *** ****** * number ** ***** *************** including:

**** ********* * (**** Dynamic ***)
**** *********
***** ***** *.***+
****** **********
********* *.***+
********* ***** ******
******* *-****
******* ***** ****** ** Tested (** **** ** H.265 **. *.***)
******* ***** ****** ******

**** *****, ** **** found * *** *** points:

***** * ***** ******** had *** ******* ******, with **%+ ******* ******, simply ******* **** ************ scenes *** ****** *** long **** *******, ******** great ******* **** **** I ******.
******* ***** *********** *** the **** ******* ******, though ** **** ***** bandwidth *******.
***** ***** *********** *** the ***** ******, ***** at **** **-**% *******, because ** ****** ** what ***** ** ******.

Bandwidth ***** ** **** ****** ******

****** ***** ****** *** greatly ****** ********** *******, ***** ****** ****** **** calculating ********* ***** ** high ****** ****** *** see ****** ** ** savings **** ***** ******** codecs. *** *******, ** a **** ****** ******* scene, ******** ** ****, Panasonic, *** ********* ******* were **** ***** **** with ***** ****** ******** during ******* ** **** motion, *.*., ***** *******:

*******, ** *******, ********** for **** ***** ******* and **** ******* ******* across ******* *******, ******* were ***** ** **% on *** *******.

******* ***** ****** *** have ****** ****** ** high ****** *******, ********* needs **** ** ********** for ***** **** ********,*** *******, ******** ***** ***** savings. ******* ** ** so ***** ************ *** drop ******** ** ***** links ** ******** *******.

Smart ***** *******

** ****, ****** *** major ************* ******* ***** codecs ** ** ***** some ** ***** *******, with **** ************ **** in *** ******** ** their ****. ***** ******* them, ** ***** ***** techniques ** ****** ********* (such ** ********* *** or ******** ***** *****) will ** ** * significant ************ ****** *******.

******* ** **** **** availability, ********** ***** *** surged. ** *** *******, integrators ********* **** **%+ of *** ******* *** smart ******:

*** *** *********** ***** ***** ********** 2019*** *******.

Comments (19)

Undisclosed #1

07/23/15 12:25pm

What’s the impact of new smart CODECs on existing surveillance systems? Is the enhanced compression transparent to existing VMSs?

John Honovich

IPVM | In reply to Undisclosed #1 | 07/23/15 01:19pm

We have not seen any issues integrating with VMSes to date. The changes are primarily on the encoding / camera side with decoding remaining the same for the VMS.

John Honovich

IPVM | 09/28/15 07:44pm

Panasonic is now announcing 'smart coding technology'. Here is how they describe it:

"Smart Coding Technology that modifies the encoding compression algorithm in standard H.264 video streams, yielding up to 70 percent improvement in bandwidth reduction to reduce bandwidth has been released.
Available as a free firmware upgrade to all new Panasonic 3, 5 and 6 Series Models, the technology offers significant decreased overall bandwidth usage to effectively lessen the amount of storage needed for recorded images.
Smart Coding technology encoding deploys Group of Picture (GOP) Control; Frequency Divided Filter (FDF); and 3-D Multi-Process Noise Reduction (3D-MNR) to achieve a lower bit rate for recorded images without degrading the captured and transmitted video stream. Combined, these features can effectively provide a substantial bit rate reduction (the number of bits per second that can be transmitted along a digital network) in the recorded video."

We expect to see more and more manufacturers to do similar as the bandwidth benefits are obvious and significant.

Undisclosed #2

08/24/16 02:51pm

Undiclosed Manufactuter brings up an interesting point here about switching between streams with a high GOP ratio:

To decode the current frame, the Client UI would have to decode the previous I-frame and all the subsequent p/b frames, and if the gap between I-frames is 5 to 8 seconds, this could be between 50 to 200 frames depending on the frame rate. I doubt the VMS would send these previous frames to reduce the delay, and will just expect the client to wait. Keep in mind that traditionally, I-frames are every second or two so the wait is not normally that long. Again, smart codecs can cause the delay to be much longer in some cases - it is the price you pay for increased storage...

When using a VMS to switch between streams have you noticed any unusually long delays? If UM is right, one might be waiting 20 seconds before seeing a live picture, which might be considered unacceptable by many.

Undisclosed Manufacturer #3

In reply to Undisclosed #2 | 08/24/16 04:01pm

Some Cameras & VMS allow for I-Frame insertion request to be sent to demand an I-Frame immediately. This can mitigate waiting 200+ frames for the next regularly scheduled I-Frame.

Undisclosed #2

In reply to Undisclosed Manufacturer #3 | 08/24/16 08:48pm

Is this an out-of-band request that operates on an existing RTSP stream, or is this made when creating a new socket?

Undisclosed Integrator #4

12/21/16 01:53am

We had a site where we installed a brand new Q6000 + Q60 PTZ and decided to try Dynamic GOP on the VMS. We leaned the hard way that VMD does not work when your I-frame interval is too high, we had an incident and no video because there where not enough frames to compare for VMD to kick off recording.

Dynamic GOP is great for continuous recording but be very careful on motion based recording.

Ethan Ace

IPVM | In reply to Undisclosed Integrator #4 | 12/21/16 02:08am

That's a terrible side effect! What VMS were you using? Server side or camera side VMD?

Undisclosed Integrator #4

In reply to Ethan Ace | 12/21/16 02:37am

its one of the Top 5 VMS's, Cam side VMD.

Undisclosed #2

In reply to Undisclosed Integrator #4 | 12/21/16 02:43am

If its Cam side VMD then it's not the VMSes fault, right?

Undisclosed Integrator #4

In reply to Undisclosed #2 | 12/21/16 02:52am

It's hard to say. Because the cam is doing the detection but the VMS is getting the metadata to tell it to records. It still needs to get enough frames to compare.

The more it's discussed the more questions arise on the technicalities.

I am trying bench test a similar cam but have. It had a chance to do so.

Undisclosed #2

In reply to Undisclosed Integrator #4 | 12/21/16 03:01am

Its hard to say. Because the cam is doing the detection but the VMS is getting the metadata to tell it to records. It still needs to get enough frames to compare.

so maybe the I-frame associated with the P-frames that have motion in them has already been discarded by the VMS when the meta-data comes in? What was the pre-record amount set to?

Maybe you can set that to a large enough value to cover the GOP length.

Undisclosed Integrator #4

In reply to Undisclosed #2 | 12/21/16 03:05am

We believe that is precisely what happened. GOP length was get to ~300.

I think pre/Post was set to 15 Seconds

Undisclosed End User #5

05/20/18 03:59pm

Good write-up.

Undisclosed End User #5

05/27/18 02:50pm

A great write-up.

Undisclosed #6

11/19/18 11:28am

Good to know about the CODEC guide.

Juan Banda

IPVMU Certified | 07/30/19 05:58am

The Surveillance Codec Guide link mentioned above in this page doesn't work.

Sorry, but the page you were trying to view does not exist.

It looks like this was the result of either:

a mistyped address
an out-of-date link

John Scanlan

IPVM | IPVMU Certified | In reply to Juan Banda | 07/30/19 01:16pm

Thanks for the heads up Juan. The link to the Surveillance Codec Guide is now working.

Juan Banda

IPVMU Certified | 07/30/19 06:16am

I highly enjoyed the Smart Codec Guide material presented here.

Read this IPVM report for free.

This article is part of IPVM's 6,545 reports, 882 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.