Sighthound, SecuritySpy, EvoCam

[nunuvyurbiz]

My foray into home IP cameras (5+ years) certainly has been a (technological) adventure, which is part of the appeal. I hope it doesn't violate the forum rules to comment about other products.

I figure there are three main apps for Macs: Sighthound, Evocam, and SecuritySpy. Each has advantages and disadvantages.

SecuritySpy
SecuritySpy is old school: very configurable (within its feature set), very fast, built in web server, built in SSL for the web server. But it's kind of dumb, at least compared with Sighthound. That is, it uses the traditional motion detection by detecting changed pixels This results in a lot of false positives. Someone walking outside my area of interest but with a shadow in my detection area will trigger it. It can only save to a single location, but Hazel can, for example, address that by moving, say, "MD" files to another location. While you can have multiple schedules (e.g., one per camera), the schedule is only active vs. inactive. When inactive, it won't continuously record but will capture motion if you set it so. When active, it can record both continuously and when motion is detected. That's it. So you can't, for example, capture continuously and not capture motion during certain periods (I want that - for example, capture continuously 24/7, but don't capture motion when I'm normally home.) Compare that with Sighthound's rules schedule - I can turn off person detection during evenings and weekends.

It's light on processor usage, relatively. As others have noted, it's more expensive than Sighthound as the cameras increase, but less than Sighthound with only a few cameras.

EvoCam
EvoCam is cheap at $30 with no camera restrictions. EvoCam uses an innovative approach of having the user create canvases, on which the user can put any combination of video feeds, clocks, text boxes, sensors (motion detection areas), etc. Each item can be highly customized, and can even overlap. Thus, each camera provides a single video feed into the app, and then gets painted onto different canvases for different purposes. You could have a canvas that aggregates all video feeds and use that to record, in the background, as a single video recording capaturing all cameras. Another canvas to match iPhone resolution, for remote viewing via the built in web server, also in the background. Another canvas for onscreen viewing. Another canvas for external cameras. Another one for internal cameras. Etc. But it's slow - it eats CPU, and there's a pretty significant lag between motion and seeing it onscreen. And the detection is the classic motion detection. There is a built in web server but there's no SSL support built in (and try as I might, I just couldn't get it to work with routing traffic through Apache, which I was able to do with SecuritySpy before it added SSL support built in).

Sighthound
Sighthound is intelligent - the person and object detection, while not 100%, is definitely more effective than classic motion detection. And I like the rules that allow for crossing lines, entering/exiting, center mass vs. bottom, etc., etc. And, rules can have their own active periods. You can save motion clips to a different location (on a per rule basis even), which is nice. (I'm big on this because I want my motion clips to be saved elsewhere that gets synced to the cloud.) I also like how it flags the video as opposed to creating separate files (though you can, and I do, save certain detections as separate files). But it lacks other features. You can't even tell it where to put the clock, let alone have a precision clock. That's not so bad but there's no web server (let alone a web server with SSL) and that, to me, is bad. You can't configure framerate, or compression. It doesn't support audio from cameras. You can't view the live feeds while viewing clips, and you can't view more than one live view at a time (at least, not at full framerate). The single biggest thing for me is the lack of a web server. It would seem to me that adding features would be easier than coming up with the intelligence behind the software, but after years of folks asking about the web server (among other features), I've lost a bit of hope.

In the end, I have both Sighthound and SecuritySpy running. Yeah, that's right, I bought licenses to both. I use Sighthound for the majority of the effort - recording continuously, using rules and schedules, and capturing clips when rules get triggered. (Oh, by the way, I also use Hazel a lot to clean out old clips.)

And then I have SecuritySpy running also, just to provide live web access to the cameras. It's pulling a different feed from the cameras, of course, at a lower resolution and framerate.

Since my Mac is hooked up to my TV, lately I've been toying with how to best setup logins for all this. I was running everything off of one login (everything, including all my personal files), but lately I've migrated my server-like functions into a separate login (iTunes, SecuritySpy, SIghthound). I may go further and split off the cameras, so that I have a login for iTunes/web that anyone can use without access to my personal files or the cameras. But I would like to be able to easily see one or more camera feeds while I"m in my personal login. I suppose right now I could do that by using SecuritySpy's web access from just a different login (as opposed to remotely). If anyone has experience with the best approach for this (logins on your only computer hooked up to your TV, where you want to address keeping your personal files secure, keep the camera software secure, and yet, for media, give anyone access), I'm all ears.

[303]

I am not sure if you are actually asking a question or just posting thoughts, but i'll try to respond.

I run both SecuritySpy and Sighthound. I wish I didn't have to, and could rely on Sighthound only. However I cannot until:

Sighthound...
- has a smaller memory / cpu footprint. right now it is 2x that of securityspy

- allows official api or scripting of camera operation. i use IR sensors to trigger recordings as well.

- emails me animated gifs or sequences of images instead of just a single image from a recording

I think that sighthound is sitting in a weird limbo right now. Dropcam's existence puts sighthound permanently in the prosumer / home security enthusiast category. Dropcam is simply easier for most people. So then it's all us who configure cameras and can deal with the techier side who turn to solutions like SecuritySpy and Sighthound. The problem is that Sighthound has superior usability, computer vision and user interface, but actually more limited tweakability and integration opportunities. IFTTT integration is a great first step and I'm really looking forward to seeing where they go with future improvements.

[nunuvyurbiz]

Just posting thoughts. I take "discussion" forum literally. Though there was a question at the end about setting up accounts.

Thanks for sharing!

[303]

So is this still your setup? I am trying to do the same, but pulling two streams gets pretty bandwidth intensive

[nunuvyurbiz]

So is this still your setup? I am trying to do the same, but pulling two streams gets pretty bandwidth intensive

True. Playing around with it just now I managed to get Sighthound to pull in streams from the SecuritySpy web server. I haven't analyzed the impact on bandwidth, ability of Sighthound to get 10fps, etc., etc. yet though.

[303]

What I'd ideally like is to just point sighthound locally to securityspy published streams on localhost so that only 1x the bandwidth is being used

[nunuvyurbiz]

What I'd ideally like is to just point sighthound locally to securityspy published streams on localhost so that only 1x the bandwidth is being used

That's what I did.

[nunuvyurbiz]

Since the above messages, I've switched to wiring all my cameras that can be wired because of their location. This has, of course, greatly reduced the load on my wifi network. As a result, I also increased the frame rate, from the cameras, to the maximum. I still have Sighthound connecting to and pulling in the feed from the SecuritySpy web server.

I was asked privately how I did that. Suffice to say, select an all manual configuration:

- "Network IP camera"
- "other"
- "manually specify the address..."
- "other IP camera"
- "http"

and then use 127.0.0.1 as the IP address, add the port SecuritySpy is listening on, and add the URL for the camera based on the URL indicated when viewing only one camera in a web browser connecting to SecuritySpy.

- 127.0.0.1 and port 8000 (if using the default SecuritySpy http port)
- and then the stream path using the URL

For me, then, the combination is:

http://127.0.0.1:800/++video?cameraNum= ... =906&26549

[303]

@sighthound folks, any chance you can weight in here on what the best method for a local stream might be?

[nunuvyurbiz]

Just a quick update on this. While this was working fine, recently I switched to (1) wiring up the two cameras that Sighthound was tapping into via the SecuritySpy web server, and (2) switching Sighthound to now point to the cameras directly. I figure that since they're wired, there's plenty of bandwidth. This has reduced the CPU load on the machine hosting all this.