*********, *** ******** ******** ***** '** is'. ***** ***** *** ** ***** brightness, ******** *** ********** ********, ***** require ****** ********** **** *********** *********.
** ********, ********'* *** *** ** interesting ******* ****** '******* ***********' **** automatically ******* *** ******* ** ***** displayed. **** ** **** *** ******* looks ****:
**** *****, **** ******* ******* **** up, *** ******** ************ **-******* ** its ***. ************, ** ***** ** any ****** ********* ** ********'* ***, not **** ***** ***.
*** ********* ******* ********** ********* ************ in * **** ******* ** *******, including ***** ***** **** ***** *.***. This ** * ****** *** *******.
******** ******** ********:
****: ** ******** *** **** ******'* image ******* ** ********'* *** ****** the ******'* *** *** ******* ******* (for *** '******' *******) *** ***** VMSes (**** ** *****). ***** *** camera *** ******* *** ***** ***** displayed *** **** ***** *******, ********'* regularly ********.
Display ********** ** ******
** **** **********, ** **** ******** of ********'* ***** ********** *** *** effects ** ********* *******. ** ******* each ******'* *********, *** *********** *** effects ** ***** ********** *** ***** and ***** ****** ** ******** ******, showing ***** *** *** ******** ******.
Follows ******* ****
***** ********** ******* ******* ****, ** only *** ******* ******* ** *** image ** ***** **** ******* ** the *********, **** ********** ********* ***********. Though ***** ** *** ********** *****, this ******* ** ****** ************ ** the ********* *******. ***** * ****** with *****-** **, **** ******* ** the ****** *****, *** ***'* ******* plate ******* ** ** ****** *** and **********:
*******, ******* ** ** **** *** plate, ** *** *** **** *** image ** ******** ***********, ******* * much ******** *********, ****** *** ***** easily ******* (*** ********** ****** ************* for *******):
***** *** **** ******** **-****** **** ** **** ******* *** ***** ********** *** ********** (requires******** ******).
Low ***** ***********
** ***** ******** *** **** ****** on *** ***** ******, ****** ***** effects **** ********* ** ******* **** performing *** ***** ******* (*** ** higher **** ***/** **** ********* *******). Since ***** *** ***** ******* ******* a **** ****** ****** *****, ***** is **** ******. ****** ****** ***** more **********, ** **** **** *** be ******* ** ******* ******** ** the *****. **** ********** ************ ***** effects:
*** ******* **********, ** **** *** 720p ** *******, *** *** **** color ****, *** *** **** *** known *** *** ***** ***********. **** the ***** **** ******, ********'* ***** adjustments ******* * **** ******** *****, with ******* **** ****** *******. **** did *** *********** ********* ** ********* detail, ** *** ***** ******* **********, and *** ******* **************. *******, ** is **** ****** *** *********.
******* ** *** **** ***** ** Avigilon, ** *** *** ** ** bright ******** ** *** *** *********, though **** ********** *************, ******* ****** yellow/green:
** ********, *** *********** ***** ***** the *** ***** ****** *** *******. Again, *** ***** ** ********, ****** detail ** *******. **** *** ** of ********* *** ** ***** *** camera *** ********* ****, *** ** limited **** ******* *** **************.
Evenly *** ***********
*****, ** ****** *** ****** **** camera ** **** *****, ******* ***-*** lx. ******** ******** * ********** ******** image, ****** **** **** *** ********* to ** ******** ** *******. ********'* gamma *** *****/***** ***** ********** ******* in ****** ************, ***** ******** ****** on *** ***** *** *******'* ****.
****** ***** ** *** ****-*** ******, in **** *****, ***** ** ****** difference ** ***** ** **********, *** we *** ******** ******* ** *** camera *****, ********* ** * ****** loss ** ******.
Wide ******* *****
** ****** *** **** * *** WDR ******, ** ****, ** *** if ***** ***** ********** ******** ******** wide ******* ***********. *** ***** ********** effect *** * ******* ******* ******* as *** *** ***** ** ******* image ********** ** *** ******. **** can ** **** ** *** ***** below ** * **** ***** ******* throughout *** *****, ***** ***** ** be ****** ** ***** ********** ********* on *** ***, ** ***** ******** in *** *****.
**** * ******* ** ***** ** the *****, ** *** *** **** ACC ******** * ****** *********** ** WDR ***********. *** ***** ** ******* and ******** ****** ** **** ** lower *****.
Outdoor ***********
** ***** *** ******* ** *** outdoor ******* ****** ** *** **** effect *** ***, ** ****.
***** ******
*****, ** ****** **** * ******* generation ***** ******. ***** *** ****** difference ******* *** ***.
** **** ****** ***** *********** **** an ***** ******** ********** ******. ****** were ******** ***** ** ***, *** no ***** ***** ****** *** ********.
Thermal ****** **********
*******, ** ****** *** **** * thermal ****** ** *** ** ***** were *** *********** ** **********. ******, Avigilon ********* ********* ** **** ******* in *** ****, **** ** *** plant ** *** ******, *** ******* the ***** ** ***** ******* ****** against *** **********. **** *** ********* to * ******** ****** ** ********* when ***** ******* *******, ***** **** contrast ** ***. *******, *** ***** the ******* *** ******* ****** ****** lost ** ***** ** *** *** to ************, *** *** ***** ** his **** **** **** *******. **** could **** ** **** ********* ** identify ******* ** ****** ** ***** in *** *****, *** *******.
1. Can 'Display Adjustment' be turned off? If it can (or could be), would you recommend only turning it on in low light, based on the testing above?
2. Could a similar feature work with higher compression methods like MPEG or H.264?
3. How does 'Diplay Adjustment' affect images from low-res analog cameras via encoders?
I'll let Ethan answer the specific questions. One thing to clarify - what do you mean by 'higher compression methods'?
Sorry, bad wording.
I meant compression methods that produce smaller file sizes like MPEG, or further, H.264. I am assuming you tested using JPEG2000?
Undisclosed, your answers:
1. It can't directly be turned off (no simple checkbox), but you could set black level all the way down and white level all the way up, which would effectively be the same thing.
2. Cameras tested were 100% H.264, and almost all non-Avigilon. To be particular, the license plate shot was an Avigilon bullet camera. That was the only one.
3. Good question. I haven't directly tested it, but I imagine it may be similar to the last outdoor shot above (cars in the parking lot). That camera was a VGA IP camera. It's not going to give you tons of extra detail (or much at all, really), but it may fix the contrast a bit and increase visibility ever so slightly.
To clarify, most (or all) of these tests were done using H.264 streams including many from non Avigilon cameras. This is independent on JPEG2000, though I know we all love JPEG2000 :)
Would you expect the results of 'Display Adjustment' to be better or worse (or differ at all) using JPEG2000 or MJPEG?
I'd think the results wouldn't differ at all. I just changed a couple cameras to MJPEG, and there was no noticeable difference.
This function is not new and not exclusive to Avigilon. NICE has had this feature available since their NET2.0 release. It is called Visual Parameter Optimizer or VPO and it is included at no charge with the NICE Enterprise Level software.
Thanks for sharing. Here's a video demonstration of NICE's VPO feature and NICE's application note on it.
Btw, we did not claim it was 'exclusive' to Avigilon. The challenge with NICE, in particular, is that it has very limited distribution. Secondarily, Avigilon's option is included even in the lowest tier version.
If anyone has other VMSes to share that have similar features, please share.
Ethan, did you notice any "pumping" of the image brightness/contrast on scenes with varying light levels? It seems to me that was a defect we noticed when we evaluated an Avigilon system last year. Specifically, a camera pointed down at a gaming table rapidly changed overall table brightness as the dealer's hands moved between brighter and darker areas.
I'm assuming all this processing is done client side at the time of viewing? You certainly wouldn't want the VMS enhancing the stream before saving to disk. Better to work with the raw stream every time.
I thought I saw a Genetec guy demoing something similar at ISCW. I was passing by so I could be wrong, but this is what it looked like. He certainly wasn't showing results as dramatic as those pictured here.
This can be done fairly easy on video exported from any VMS with some basic video editing software or a few VLC plugins. The settings being manipulated are fairly basic (which begs the question - Why don't far more VMS's do this?). They have made a clean interface that will certainly reduce the steps needed to pull that off. That the image processing is done on only the zoomed area is genius.
James, yes, it's client side, and does not change / manipulate the original video.
And it does seem something that many other VMSes could and should do.
I think you should take them seriously and believe that most of Avigilon's dealers are happy with them (and this is coming from someone Avigilon dealers accuse of being biased against them). Beyond which, if you are a dealer, the marketing only helps them sell :)
Here's some comments / debates among Avigilon dealers / ex-dealers.
Histogram equilization is a neat idea. Sony uses this as well in some of their cameras. I'd like to see a clipping function added to this so that you know where and when certain areas are washed out and have some control.
