Testing New Dropcam Pro Camera

IPVM

A few integrators have asked us about whether they should use Dropcam.

The answer is clearly NO, unless it's for home/personal use. Dropcam today is the type of product that is safe to recommend to a friend or family member because it works well and it is unlikely to force you to go their house and work on it.

Dropcam is a closed / proprietary, small camera count offering that does not work with any regular professional (Milestone, Exacq, etc.) nor budget (Dahua, QSee, Swann, etc., etc.).

Moreover, Dropcam is no threat to the end user today buying 8 or 16 camera systems (or obviously larger). But I would be stunned if Dropcam did not expand to larger camera counts in the next few years. And with their money and what they have shown they can do already, I think they will be a force.

Ross Vander Klok

IPVMU Certified

You need to stop showing outdoor pictures until it is summer. 30 degrees in Michigan..... I am going to get one of these to try at home. I am thinking if they continue to expand their offerings this is something I could use with the execs because it is easy and straight forward.

Jim VanVorst

A couple more differences on the Pro:

It will do 1080p (says on the box).
It will do 5Ghz wifi (802.11 b/g/n).

As far as the 'enhance' feature, I suspect their sensor is larger than 1080p and they crop to 1:1 at the sensor to increase the resolution at the expense of FOV. This is nice to have with such a wide angle lens since there is no way to optical zoom on a region of interest. The penalty is that you are not recording anything outside of this 'enhanced' region.

As-is, Dropcam will never expand to more then a small handful of cameras at a given install site, since each camera is permanently taking up a large chunk of your upstream bandwidth. Also, if you full-screen your video you will see that the image quality is awful, since it has to use a low bitrate. It looks fine on small screens (mobile devices) and as a small window inside a browser.

John Honovich

IPVM

In reply to Jim VanVorst

Jim, great feedback.

The 1080p thing is a little 'weird'. It definitely streams at 720p, not 1080p. We exported many video clips and used a stream analyzer and it was 720p only. I've seen a couple other sites say things like "The Pro currently shoots video in 720p, but company reps said that it's capable of 1080p."

Actually, to that end, I suspect the sensor is 1080p but they are only streaming out 720p which lets them get higher image quality when they zoom / enhance.

And yes to the lower image quality. They use a higher than average quantization level, something we covered in the original test.

Finally, I agree that as is, Dropcam will never more than a small handful of cameras. On the other hand, it would not take much work to embed a 32GB (or 64GB SD card) inside a future Dropcam 'Edge' and overcome the issue that way.

Ryan Gomes

Sounds like Dropcam will have to disrupt their own business model if they hope to expand into the 4-24 camera market. If they store video at the edge, they won't be able to justify the recurring monthly cloud storage fee.

John Honovich

IPVM

In reply to Ryan Gomes

Yes but they would also reduce their storage / bandwidth costs. So instead of $9.99, charge $3.99, still RMR but lower revenue offset by lower costs.

The bigger threat I see to Dropcam is that traditional hardware manufacturers embed storage and then give the remote access away for free, undermining Dropcam's model.

Chris Dearing

•Nov 29, 2013

In reply to Chris Dearing

"Enhanced: Dropcam has added an 'enhanced' image feature that is pretty amazing, very useful and something that we have not found in any 'real' IP camera."

How is this any different than Panasonic's VIQS?

Except for the fact that you can see the enhanced and the unenhanced at the same time. Maybe that's why they use the bigger sensor, but don't actually stream it?

John Honovich

•Nov 29, 2013

IPVM

Chris, great question.

What Panasonic doing is varying the quantization/compression level in different parts of the scene. The 'high' part might get a q=27, the low part might get q=32, etc. As the doc you linked to shows, it's really about modest bandwidth/storage savings from modest quality losses.

By contrast, with Dropcam enhance, the bandwidth stays roughly the same whether you do the full area or an 'enhanced' region. What changes is (1) how many pixels are allocated to a region (done evidently because the sensor pixel count is higher than the max stream size - 1080 vs 720) and (2) the exposure (which can be optimized for the smaller area being 'enhanced'.

We've tested VQIPS in the past and found its benefits to be modest. The Dropcam enhance feature has a lot more upside.

Chris Dearing

That seems like a fair comparison between the two, however it still seems to me that on a technical level they are doing the same thing.

Kindly show the error in my thinking or my assumptions:

Assumptions

The camera does not physically do anything to go into enhanced mode.

The camera has one 1920x1080 CMOS sensor.

The FOV captured by the sensor in standard mode is the same FOV captured in enhanced mode

In standard mode the FOV shown on the monitor is the same as the one captured by the sensor

In standard and enhanced mode the data stream is that of a 720p stream, which is not the native resolution of the sensor

Standard mode must trans-size(compress) the 1080p stream to a 720p stream by combining the pixels in some (lossy)way

So the dropcam trans-sizes (compresses) the whole image in standard mode from 1080p to 720p and in enhanced mode trans-sizes the image less depending on the desired FOV

Essentially the Panasonic is doing the same thing its just seems different because the full sensor is available at all times, so when you turn vqis on your quality decreases, but with the dropcam the finest level of sensor detail is only available in enhanced mode so your quality increases.

Dropcam is most likely throwing away some of the image quality to decrease the size of the stream to their cloud. Otherwise they would just give u the full sensor stream all the time and their would be no need for a 'magic wand'

Don't agree?

Pray tell what exactly what is the thingy of which you speak that determines how many "pixels are allocated to a region"?

impressive marketing for sure

John Honovich

In reply to Chris Dearing

IPVM

No, the technical implementation is clearly different.

With Panasonic, VQIPS is always streaming the full fov, the only difference is what parts gets a higher or lower compression level.

With Dropcam, enhanced mode only encodes a subsection of the FoV so part of the FoV is thrown out and the other 'enhanced' part gets encoded.

The reason why the image quality is significantly better with dropcam 'enhanced' is that (1) they adjust the exposure just for the specific region (panasonic VQIPS does not adjust exposure and (2) since Dropcam maxes out at a 720p stream but uses a 1080p imager, what they are essentially doing is digitally cropping the 1080p stream to the 'enhanced region' to a 720p stream.

The implementations are completely different.

Chris Dearing

Thanks for the detailed response...

Ok, I totally agree the exposure handling should be benificial.

I also would admit that my statement 'technically they are doing the same thing' is clumsy and vauge, although I never claimed the 'implementations' were the same.

As for the resolution issue, I also agree with you that they are digitally cropping the 1080p to 720p in enhanced mode, however...

If they are digitally cropping in enhanced mode, what do you think they are doing in standard mode?

Digitally compressing(transsizing)right?

Perhaps you don't consider transsizing compression, but in the case when u go from higher to lower resolution its effect is the same: irreversibly reducing the size of the byte stream

That's what I meant by the "same".

No doubt the area of interest looks better in dropcam enhanced mode than standard mode.

What I am trying to say is the only reason it looks better is because you don't have access to the whole 1080p sensor. If you did the magic wand would lose its magic.

You said that the VQIPS is really about 'modest bandwidth/storage savings from modest quality losses'

I agree!

But when you look at the reason the dropcam is digitally transsizing from 1080 to 720, its for the same reason, to save bandwidth, god knows its not for quality!

Maybe all mfrs should make their sensors bigger than their stream, so they can say "these go to eleven!"

If we were both given identical Ferraris but yours had a governer that limited its speed to 2/3 of its actual top speed, but had a big red button that you could hit twice an hour to give you a five minute boost to top speed, would you reckon yourself better off than me?

Do you see at all what I'm saying, or am I not making any sense?

John Honovich

In reply to Chris Dearing

IPVM

First of all, on a technical level they are NOT doing the same thing.

Dropcam adjusts exposure. Panasonic does not.
Dropcam eliminates a portion of the FoV. Panasonic does not.
Dropcam keeps quantization constant. Panasonic varies it (based on the high and low zones).

What you seem to be arguing towards the latter end is that Dropcam's approach is pointless or counterproductive.

However, for Dropcam's business model and customer segment it is quite useful, more impactful than VQIPS is for Panasonic's.

With residential hosted video (Dropcam's focus), there is not a lot of need to go from 720p to 1080p, plus the increased upstream bandwidth requirements and higher cloud storage costs are negatives as well. Instead of going 1080p by default, using a higher pixel count imager, lower default stream and this enhanced focus adjustment / digital cropping, they can sell a single wider angle camera but make it useful for people who only want smaller FoVs.

Chris Dearing

Ok, I acquiesce!

I concede that they are more different then they are similar in general, and in addition from a interface point of view they are different in that dropcam allows ad-hoc looking here and there real-time, and with the panasonic you are setting up more or less permanent zones.

In any case according to Dropcam corporate(as you allude to above), tbe limitation is not in the camera but due to limited back-haul capacity (my paraphrase) , which they are fixing as i type.

So then the wand becomes a magnifying glass (when streaming 1080p)?

Finally, please allow me to save face by probing your, "people who only want smaller FoVs"... statement further. Might these people (depending on the crop) be just as well served by the dropcam hd?

Mark Rindy

https://www.dropcam.com/c/adaeab4bd4784a9cb98a9fc4f1281410.mp4

We specify and install the standard VMS and best of breed camera products for projects big and small. While I'd like to have one of these systems at home, it is not sensible to spend $10K for a home video system.

We use the Dropcam cameras for security, shooing solicitors away, and capturing those oh so wonderful moments where the kids are showing off their new vocabulary: Warning, foul language.....from a 2 year old.

Happy Thanksgiving!

Brian Rhodes

IPVMU Certified

In reply to Mark Rindy

I think he fingered you as the root cause after mom shut him down, Mark.

Mark Rindy

In reply to Brian Rhodes

Indeed, turns out that I taught him during an afternoon rush hour 'discussion' with another driver. Forgot he was in the back seat and certain he didn't hear me but with dumbo ears like his dad, that was not the case.

Huzaifa Najmi

•Jan 23, 2014

Any idea which sensor is used in Dropcam Pro ?

Gautham Gopalakrishna

I see quite a few reviews in Amazon stating that the latency is about 10 seconds for live view. Digging a little deep, dropcam uses HLS for live streaming. While this is great for bandwidth aware playback, a definite no in enterprise scenarios where 10 second latency is not acceptable. Of course Dropcam is intended for homes, but 10 second isn't great even for homes. What do you think?

Have you observed similar latency in your tests?

John Honovich

In reply to Gautham Gopalakrishna

IPVM

Gautham, I was getting an average of 3 - 5 seconds but obviously latency depends on a number of factors so others could be getting longer. Either way, there's definitely a delay.

A few seconds latency (close to 3) is manageable for most home users in most occassions. 10 seconds would be more of a problem but I am not sure how many people are actually getting / suffering from 10 second delays.

Gautham Gopalakrishna

John,

Thanks. I saw a review in Amazon with singe * rating that had this issue of 10 seconds. Also there's another review in Apple store with 1* review that mentions 10 seconds. If you still have the camera with you, I'd appreciate if you can publish the upstream & down stream bandwidth that you have when you measured 3-5 seconds of latency.

Apple mentions the 10 second latency issues with HLS here & the recommended settings even shows 30 seconds (in the same page, question 21 in FAQ)

Ross Vander Klok