Verkada Video Quality Problems Tested

By Ethan Ace, Published May 23, 2019, 10:53am EDT

Verkada suffers from numerous video quality problems, not found in commercial IP cameras, new IPVM testing of Verkada vs Axis and Hikvision shows.

free image_3

Fundamentally, the problems originate from Verkada using a decade-old approach to streaming. instead of a modern VBR/MBR, smart codec implementation.

Problems include:

  • Heavy nighttime detail loss compared to competitors in low light
  • Significant detail loss in daytime scenes compared to competitors
  • Reduced detail in WDR scenes due to artifacts on moving objects
  • Higher bandwidth consumption than competitive cameras
  • Live and playback stream buffering issues
  • No user adjustable compression or image quality controls

We explain each one inside with images and videos demonstrating the issues.

*******

** *** *****, *******'* cameras ********* ********* *** to **** ****** ************/*********** levels **** ******* *******, with ******** ** **-** common, *********** ************ ** ********* 28-30. ******* ******** ** user-accessible ******* ******** *** this.

******* ** **** **** compression **** **** ******* at *****, **** ******* details *********** ******* *** facial ******** ******** ** artifacts ** **** ******.

*******'* ******* ******** *** much ****** **** *********** in *** ****** ******, with ******** ************ ** the ***** ** *,***-*,*** Kb/s, ********** ** ***** complexity, ***** *********** ** quality ******** *** ********* compression ** **** ** making ***-**** ***** ************ more *********.

************, *** ******'* *** negatively ******** ***** *******, resulting ** ******** *** doubled ***** ** *******, as **** ** * slight ****/*** ****. ** with ***********, ***** ** no ****** **** ********** means ** *********** *** levels.

*******, ****** *******, ** experienced *********** ********* ******, with **** **** ** SD *** ** ******* frequently ******* *** ******* ahead, ********* ** ********** missed ******** ** *** scene ** ******** ********* (though ******** ******* *** not ******* ****).

Verkada ********

** ****** *** **** results **** *******, ** which **** *******:

****** *** ********** *** deeper ******* *** *******. Your ******** *** ** line **** *** **************.

Future ************ ********

***** **** **** *** shared ***** ****** *****, it ** ******** **** Verkada ***** *** **** of ***** ****** ** future ****** ********. ***** other ************* **** ****** these ******** *** *****, it ** ******** ******* will ** **, ** well.

No **** ******* ** ***********

****** *********** ***** "**********" camera ******* ******** ****, Verkada ******** ** ***********, quality, ** ********* ********. The **** ****** ** the "*********" **** ** their ******* ** ********/********* Cloud ******:

Verkada Bandwidth Menu - Cloud Archiving Settings Only No Bitrate Controls

********, ***** **** *********** use ***, **** ****** variation ** *******, ******* cameras' ******* ****** **** significantly. *** ** ***** issues, ***** ******* (**** as **** *** *********) fix *********** *** **** the *******, ***** ******* primarily ****** ***********.

High *********** ** *** ******

********* ******* ****** ***** from ******** ****** ******* from **** *** *** motion ** ******* ********* parking ****, ** ***** that******************** ****** **** ******* in ************. ******* ******* ranged **** ** ******* of ***** ** ** 38 ************ (** * scale ** *-**), ******** to **-** *** ******* standardized ******* (***** **** less ***********, ****** ** ******* ***** Quality / *********** ******).

**** ******* ** ******* details ******** *** *******, as *******'* ******** *** chart **** *** **** clear *** ******** ******* due ** *********, ***** below. *******'* ******* ************ was * ********** **** ~32, *********** *********, ***** Axis *** ********* ******** 28/29 (**** ***** ********* less ********* **** *******).

Nighttime ******* ********** ******* *** ** ***********

***** ******* **** **** worse ** *****, **** the *******'* **** ******* obscured ** *********** *********, with ************ ********* **, compared ** **/** ** Axis *** *********. **** the ***** *** ***** of **** *** *******, while *** **** *** Hikvision ****** **** ***** recognizable ****** ******* *** 4-5 ***** ** **** (again ***** ********* **** bandwidth **** *******).

Heavy Compression Artifacts Obscure Facial Detail

Not **** *** / ******* *** *********** **** ****

******* *********** ******** ** our **** ****** ** a ******* **% *****, averaging ***** * **/*. By ********, ************ ******* in **** **** ******* more *************, ********* ********** by***% ** ****.

******* ** ******** *****, Verkada ********* *** ******** lower ** **** ******, such ** ******* *******, though **** ***** ******** higher, **** ** **** indoor ***** *** ******* parking *** *****.

Bitrates Consistent In All Scenes

**** **** ***** ************ are ***** ** ******** clips. **** ****** ********* was ******** ** ** ~5-10% ***** **** ******** in *** ****** ******, though ** ****** ** directly ******** *** ********, as **** ****** ** available **** ** ***** web ** ****** ***.

Higher ********* **** ***********

******* ** ***** *********** bitrates, ******* ********* *** higher **** ****** ** all ****** ******, **** Axis *** ********* **** more **** **% ***** than ******* ****** *** day, *** ***-*,*** **/* lower ** *****.

Bitrate Comparison

Significant *** ********* ** ****** ********

** *** *****, ****** WDR ********* **** *******, with ********, ********, *** doubling ** ****** *****, shown *****, *** **** in *********** ******.

Significant WDR Artifacts

**** ******** **** *****, WDR *********** *** *********** with **** *** *********, shown *****, ****** ***** subjects ** ******* *** scenes *** ********, ** entrances, *****, *** ***** man ***** *** *** most ****** ************.

No ***** ******* ********

******* **** **** *** include ******* ***** ******* controls ***** ** ********** cameras, **** ***** *********** only **** ** ****** camera ***********. ***** ****** enable/disable *** ** ****** levels, ******* ** ************, switch ***/***** ****, ** adjust ********. ** ********, many ******* (**** ***** below) ******** *** ***** controls ** *** *** alone.

Image Quality Settings Compared

Live *** ******** ********* ******

** *** *****, ******* streams ********** ****** *** buffered ****** **** ******* and ********. **** ******* SD *** ** ******* live, *** ****** ****** and ****** ******* *********, with ******** ** *** scene "*******" ******** ** the ****** ****** ** to ****. *** **** stream ***** *** *** frequently ******* ***** ****** was *** **** / Low ******* ******, * lower **********, ****** ***********, lower ******* ******.

** *********** ***** ****** in **** *****:

******* ****** **** ******* in ********, **** ******* taking ******* ******* ** load *** ********* ********* every *** ******* **** attempting ** ** ** playback.

**** **** ***** ****** are ******* *********'* ****** *******, ** ****, *** only *** ******* ** used *** *******.

Test **********

******* ******* **** ****** using ***** ********, ** they ** *** ***** configuration ** ***** ***********, bitrate, ** ***** ********.

** *** *****, *********** were ***** **** ******** clips, ** ***** ******* are **** ***** ***** play **** ** ******** for ***********/*********** ********. *******, note **** ********* ***** from *** ** ********** and ******* ** *** tests **** ******** *********.

********* ** *******, ******* are ********** ** *.*** at *** **** ** export *** ******** ******** compatibility **** *** **** resolution ****** ***** ** compressed *** ****** *** H.265 ** *** ******. As **** *********, ***** is ** ******* ** compression *** ******* ** export.

**** *** ********* ******* were ****** ***** ******* settings **** **** *******:

  • **** ********* *** ******* (High, ******* *** ***)
  • ********* *.***+ *** *******

******* ******** *** "** to ****" (** ******* is ******):

**** *** ********* ******** versions ****:

  • **** *****-*** ********: *.**.*
  • ********* **-**********-* ********: **.*.** build ******

Comments (27)

It looks as though Verkada's cameras are stuck in a the stone age.  But hey, any press is good press, right?  6 articles on IPVM in the last month or so.  

Agree: 6
Disagree: 2
Informative
Unhelpful
Funny: 15

what’s the live latency like?

Agree
Disagree
Informative
Unhelpful
Funny

Latency is about 6-8 seconds on the SD/HD streams, and about 2 seconds on the "Live" low latency stream. From our Verkada Cloud VMS/Cameras Test:

Live Low-Latency Streaming (2-3 seconds):

SD - 300Kbps Streaming (6-8 seconds):

Agree
Disagree
Informative: 7
Unhelpful
Funny

Hey, at least they cost 50% less than the competitors. Oh wait...

I guess this is part of going direct to end-users. Some end-users have awareness; some don't. If you don't know how to use all those camera configurations, it's a no-brainer. Of course you would use Verkada because they make it simpler! Set the orientation, turn on auto-levels and go back to your normal job. Only advanced users care that all the advanced stuff is missing (or that the image is comparatively bad).

Maybe we as an industry need to think more in terms of making things simple for the end-user. We need to create simpler interfaces. We need the NVRs to run with almost no maintenance. We need to work as smoothly as Verkada.

Or maybe Verkada will just blow over. If they're simultaneously more expensive and lower quality, somebody is going to figure it out, eventually.

Agree: 8
Disagree
Informative: 4
Unhelpful
Funny

I'm hard pressed to think of an industry in the 21st century where a revolution in simplicity, even when the resulting product is objectively inferior in feature set, hasn't resulted in significant adoption. 

Simplicity can be a feature set, a business enabler, in itself. 

On a more comforting note, have there been incumbents in other industries that have appropriated the approach of the "simplicity usurper" and stamped them out? I think so.

Agree: 1
Disagree
Informative
Unhelpful
Funny

Verkada has the most aggressive sales team I've ever seen in this industry, more aggressive than Schmode Avigilon, anyone particularly agree/disagree?

That is what I think is the main difference. And I do think many buyers will care once they realize it is an issue.

Agree: 11
Disagree: 1
Informative: 1
Unhelpful: 1
Funny

IMO they have the best social media marketing campaign out there right now. This is something that really sets them apart. 

Agree: 1
Disagree
Informative: 2
Unhelpful
Funny

I agree.

I think Verkada is high pressure to try and close the sale before people realize the limitations, whereas Avigilon B.J. (before James) was high pressure to drive growth and valuation, but the product itself really didn't have severe limitation that were glossed over in the demos.

Agree: 4
Disagree: 1
Informative
Unhelpful
Funny

whereas Avigilon B.J. (before James) was high pressure to drive growth and valuation, but the product itself really didn't have severe limitation

On the other hand, Verkada is just in year 3. Avigilon was not a great product in 2007 either. So it will be key to see how much Verkada improves in the next few years. Surely, buyers will start learning their flaws so the race is whether they can fix them faster than buyers broadly recognize them.

Agree: 4
Disagree
Informative
Unhelpful
Funny

Exactly.

On a related note, I believe anyone selling "traditional" IP cameras or VMS should be very careful in selling against Verkada only or primarily on image quality / specific features, rather than going after the model / infrastructure. It'll be tempting for quarterly quota sales, but long-term these things will likely be smoothed out, and then you've got no response.

On the other hand, depending on the decision-maker, you might have to run purely on current quality.

Agree: 3
Disagree
Informative
Unhelpful
Funny

I do absolutely agree that the aggressive sales team is a major part that's driving their growth. But I also think they couldn't sell nearly as well if they didn't have a smooth interface. I mean, people don't just buy anything because the sales person is aggressive, do they? (Did I just hear somebody open a new tab to look for an applicable scene from Boiler Room?)

Maybe we should all just hire better/more sales people. But maybe there actually is something to the product that the industry hasn't understood and learned from yet. I think it's a little of both.

If it's just the sales team that's driving the growth, then they are sitting on a time bomb, and had better get ready to pivot quickly to spending a bundle on engineers to build better hardware.

Agree: 1
Disagree: 1
Informative
Unhelpful
Funny

I think the most important part of their 'tech' for their sales effort is that the cameras can get up and running without any technical knowledge because the storage is inside and it uses cloud connectivity. This empowers, if you will, people who know nothing about video surveillance to proceed with buying.

My hypothesis is that eventually, most of those people will figure out that things like lack of 3rd party support, night time video quality problems, delays / stutter in streaming video, lack of 40 various VMS features are problems. We will see.

Agree: 6
Disagree
Informative: 2
Unhelpful
Funny

My hypothesis is that eventually, most of those people will figure out that things like lack of 3rd party support, night time video quality problems, delays / stutter in streaming video, lack of 40 various VMS features are problems. We will see.

My hypothesis is that the majority of the actual security camera buyers market doesn't care about most of that stuff. And for those that do care to a moderate degree, the direct sales approach, and efficiencies of that, along with monthly billing, make up for a lot of other things they are lacking.

I think Verkada will catch up on most of the missing items faster than accounts churn for lack of these things.

There is a huge (YUUUUGE!) segment of buyers that "just want some cameras".  Buyers in this segment are often stuck between the online self-serve options, which are all kind of lacking, and the professional solutions that require they work with (and pay) integrators. Verkada is tapping that market, and I think much like Dropcam did, they have the potential to overtake a massive market segment pretty quickly.

Agree: 3
Disagree: 1
Informative: 2
Unhelpful
Funny

Verkada is tapping that market, and I think much like Dropcam did, they have the potential to overtake a massive market segment pretty quickly.

That's a good analogy though, interesting, in retrospect, as much as Dropcam made being acquired by Google, they never became a major force in the industry. Maybe that had to do with Nest screwing them, hard to say in retrospect but even in the residential security market, they were never more than a niche player overall.

My analogy, then, is even if Verkada becomes Dropcam like (I am sure the founders and investors would be happy) I am not convinced that they will be a major long term player (like an Avigilon or an Axis).

Agree: 1
Disagree
Informative: 1
Unhelpful
Funny

The majority of buyers may not, certainly not in the consumer/residential and SMB markets, however, Verkada is marketing and selling to very large enterprise customers (cities/public safety/schools).

This is the one market segment, from my experience as an integrator, that actually cares about these issues. Someone inevitably will ask: how do we connect these to our access control system? Our public transit system has hundreds of cameras on buses, trains, city vehicles, how can see these in Verkada, or what camera do you offer for use in a city bus? We need to see our Aiphone intercom cameras at the front security desk without having a separate software, how can you integrate those?

Agree: 4
Disagree
Informative: 2
Unhelpful
Funny

I've never had a vendor contact my CEO directly during a discovery period but Verkada sure did and every decision maker in between. 3 of us got unsolicited Yetis via UPS.

When I said I like the cameras and hinted at some upcoming bids the sales person responded with "we don't do bids" and "people either see the value of our cameras or don't"

Like dude we are a small government entity. We have to do bids. 

Agree: 2
Disagree
Informative: 4
Unhelpful
Funny: 6

3 of us got unsolicited Yetis via UPS.

That's hysterical! They once unsolicited sent us a bottle of champagne :)

Thanks for your first comment!

Agree
Disagree
Informative
Unhelpful
Funny

Who’s the dinosaur again?

Agree
Disagree
Informative
Unhelpful
Funny: 4

I'd put in HD-SDI (oh boy) before Verkada anyday.  Lets have Mr. Rockoff to go educate these end users.

Agree
Disagree
Informative
Unhelpful
Funny: 2

I would literally pay money to hear a debate between Todd Rockoff and Ryan Young. 

Related: The Worst Security Marketing Ever

Agree: 2
Disagree
Informative
Unhelpful
Funny: 9

Hey Ethan, were you guys able to map out where the Verkada camera destination IP(s) are routing to? Are all the models route the same? Also curious if the camera pings anything peculiar. Thanks.

Agree
Disagree
Informative
Unhelpful
Funny

They essentially only ever contact Amazon Web Services. You can find that via Wireshark, but they also show IP addresses in the cameras' Audit Log:

I haven't seen any evidence of them contacting other addresses.

Agree
Disagree
Informative: 3
Unhelpful
Funny

This seems to be a log of where request to camera originates from. This means UI is hosted on AWS. 

Assuming camera itself is not published over public IP, there should be some way for camera to setup connection to server. I think Undisclosed #7 asked about this point. For this wireshark log would help.

Agree
Disagree
Informative
Unhelpful
Funny

Great, folks. Really interesting tests and results.

Can you do the same with your Meraki cameras, please?

Agree: 5
Disagree
Informative
Unhelpful
Funny

Users complain if the latency can be measured in seconds.  If there is a single second of latency that is nearly catastrophe level complaints in my experience.  Thinking back to the mid 2000s when we were deploying IP cameras completely without any clue as to what we were doing we would see similar skipping scenarios to the above as we saturated the network or pushed the cameras frame rate to the max of what it was capable of (e.g. Axis 2100/2110/2120).  This was all school work back then and comparable to Verkada's demographic.  I cannot emphasize enough how much time was spent getting berated by the end users, particularly when we were like 7th place in the bid and were still selected.  

Eventually, with the ability to play with settings at the camera we were able to make it work and learned some pretty harsh lessons.  Looking at Verkada's 10x worse latency and skipping would give me flashbacks and panic attacks especially when compared to the competition nowadays.

 

Agree: 2
Disagree
Informative: 1
Unhelpful
Funny

it seems like as technology gets better, latency gets worse...

NTSC -> IP -> Cloud

Agree
Disagree
Informative
Unhelpful
Funny

Not technology as whole, but specific implementation of live stream.

There are different ways to provide live streams with low latency. Example would be WebRTC with less then a second delay widely used in video conference, RTMP used by flash players, RTP in browser used by Axis camera web UI, etc..

Agree
Disagree
Informative
Unhelpful
Funny
Read this IPVM report for free.

This article is part of IPVM's 7,214 reports and 961 tests and is only available to subscribers. To get a one-time preview of our work, enter your work email to access the full article.

Already a subscriber? Login here | Join now
Loading Related Reports