West Seattle Birdcam

After a brief hiatus, Birdcam 2.01 beta is back online.  A bit of soldering to repair the camera cable and installation of a 3″ ABS sleeve to protect the cable over the feeder pole will hopefully do the trick to keep those pesky squirrel-saboteurs from hiding from the all seeing eye of the birdcam.

We’ll give it a go and see how it works.  Enjoy!

Well, I said Birdcam 2.0 was a beta for a reason. Everything has been working out well with the new set up. I even thought we had the squirrel problem beat for about a week until I caught Wally (or Wanda) the Squirrel atop the feeder. So this last weekend, I made some slight adjustments raising the squirrel baffle and all seemed well.

I guess these squirrels mean business, because at some point this morning (around 10 a.m. by the last snapshot) in their frustration they managed to chew through my USB cable to the camera. This was something I was marginally concerned about initially, but I thought they would be so focused on food that they’d probably just leave the wires alone. Obviously these squirrels are professionals and not unfamiliar with revenge attacks. Not only did they chew through the USB cable, but also the zip ties holding the cable connection housing to the feeder pole. I think they’re trying to make this personal.

So, out comes the soldering iron and another trip to the home improvement store awaits so I can harden the set up against rodent assault. Stay tuned!

Decent weather and the increase in spring activity finally propelled me to get the new configuration set up. I’ll call it Birdcam 2.0 Beta because, while its working, I’m sure I may be making some adjustments and everything is only in a semi permanent state.

Here, you can see West Seattle’s finest example of birdcam engineering… The set up now consists of a pole mount feeder which has a second feeder mount attached. On it, some pvc pipefittings and an electical conduit box provide the “platform” for the camera. The camera is contained within a weather proof (I hope) housing which is just a modified outdoor lighting fixture. The housing is mounted to the base with a hinge and holes are drilled into the base to accomodate the cabling. The camera’s usb cable connects to a usb active extention cable. These active extension cables can be daisy chained to up to 80 feet I’m told. I’ll need to weather proof the cable connection between extension and camera, but that’s it.

With a very stable platform, motion detection becomes much easier. Since the pole mount feeder doesn’t swing like a pendulum like hanging feeders do, I can use a much lower threshold to detect movement and the post-detection “lock out” period can also be reduced. The net effect is that most of the shots are “keepers” and there are fewer random shots after the birds have left the feeder.

So far, it seems to be working great! Enjoy.

Spent a fair amount of time tweaking settings this weekend and looking to see if there was some software panacea available that would do all I want with the Birdcam.  I initially started out with taking snaps of the feeder at regular intervals, but that adds has a little too much of a slot machine aspect to actually seeing activity on the feeder.  So I thought I’d investigate getting motion set up.  If I got motion detection to work, I then wanted to show a recent history of motion-activated snaps of activty at the feeder.  Hopefully, some notes here may help someone else trying to set up something similar.

I’ve found that part of the challenge in running a site like this is that there are a number of wonderful widgets and interesting javascript bits available that allow you to do interesting things with webcams, slideshows and the like.  Unfortunately, most of those things cannot be imbedded in a blog post.  WordPress, where this blog is hosted, forbids the use of most, if not all, of the fun stuff because of the security threat that malicious code could pose.  I could hardly blame them, but it makes doing somethings more difficult.

After downloading and trying about six or seven different packages, I came back to ConquerCam.  It really is amazing how awkward, annoying and/or buggy so many of these are.  And that’s when their not loaded with adware.  At the end of the day, it looks like ConquerCam is quick, clean, easy and stable.  The documentation is very light (almost non-existent) but there are many helpful posts on their forum.  I discovered a few things which were helpful (and in hindsight should have been obvious but took quite a bit of poking around to confirm):

• motion detection periods can be scheduled during regular periods;
• motion detection can be limited to a user defined mask of the camera image;
• ConquerCam can keep track of and upload a series of recent history files; and
• ConquerCam will always keep the filename of the current (most recent) snap the same when using iterative file names.

The last point was key for putting these captures on a blog.  Without javascript, the current image must just point at a fixed name jpeg hosted somewhere.  With ConquerCam, I was able to have it save up to five total shots naming each capture{cnt}.jpg where {cnt} is a number between 0-4 which is inserted by the program.  The file with the “0″ in its filename is always the current image.  Mission accomplished.

Motion detection is a bit more of a challenge.  I found a very helpful post on the ConquerCam site that seemed to be key.  Everytime I turned on motion detection, it seemed to slow down or lock up the computer.   The poster suggested setting the framerate down to 1 fps if your only using the webcam to take snapshots rather than streaming video.  This dramatically reduces the load on the CPU to monitor changes to the image.

Likewise, they also suggested setting a mask for the portion of the image to monitor, further reducing the load on the CPU and making sure that extraneous activity doesn’t set off the camera.  The program comes with a few random examples of a mask, but no indication of how to actually make one.  Again, a helpful post on the forum suggested simply taking recent shot from your camera and editing it in your favorite program like Photoshop, Paint Shop Pro, or even MS Paint.  Here’s the trick:

• paint everything you don’t want to monitor absolute black (0, 0, 0);
• paint everything you want to monitor absolute white (255, 255, 255); then
• save it as a .gif file.

When you configure the motion detection, tell ConquerCam to use the mask you created.  In preview you should now see the masked area being monitored.

Detection senstivity is tough.  Two main concepts to consider here.  What’s the right triggering threshold (the amount of motion above which the camera will take a picture) and then what’s the right cooldown period (the time period after detection that the camera should not take a picture).  Just to make it obvious, here was my initial problem:  set the trigger low and when a bird came to the feeder, a picture was taken and uploaded.  Of course, as soon as it left, it detected that motion too and took a picture when the bird left, leaving the current image as a shot of where the bird just left.  Set the cooldown to a reasonable period 15-30 seconds and you can mitigate this a bit.  Its a lot of trial and error.

I’ve run into one potentially fatal flaw with my current set up though which I’ll likely address soon.  The feeder is suspended from the roof by a hook.  When birds land or take off from the feeder, they send it swinging back and forth like a pendulum which continues to set off the motion detection after the birds have left.  Same problem when its windy.  The only solution I can come up with is to pole-mount the feeder so its motion is minimized.  Another weekend project.

In the mean time, I’m experimenting with different thresholds for motion detection and it looks like its working most of the time now with the occasional false positives.  More tweaks to come!