Timelapse Experiments

I found an old camera when I was tidying up. So I decided to try and find a use for it.

The camera is a TP-Link NC250. It’s an end of life product so is unsupported. The first step was to plug it in and see if it would connect. The first experiment yielded very little. It was connecting via an extension router. I tried both WPS and ethernet, but the TPCamera app reported “Can’t Find Camera”. This is not a helpful error message.

I scanned the network to see if the camera showed up. It did show up when I connected, but it then dropped off again straight away. A few more experiments suggested that the problem may have been that the app and the camera were getting confused by being attached to a secondary router. ⊕ Something something NAT addresses something something don’t check I tried connecting to the main router and the problem went away first time.

The next issue was that the tpCamera app doesn’t have a timelapse function. The camera does motion detection and night vision, but no timelapse. I tried doing this manually, taking a picture every 5 minutes, but this gave me more of a slideshow than a timelapse. A single second of video has 24 frames. So, a picture every 5 minutes gives me 2 hours per second. That’s way more lapsing than I wanted.

In order to get the kind of effect I wanted I needed a picture every 3 seconds. There is, obviously, no way I could do that manually. A bit of googling let me to VideoVelocity. This is timelapse specific software that reads a camera feed, or a bunch of photos. ⊕ It would be more on brand if I found a way to do it in R. It looks like the RVision could have done this. I’ll try it some time. The next challenge was to find the stream from the camera. The documentation didn’t seem to mention where camera was serving the stream.

It took a bit of faffing and searching to discover that I needed to give it: http://Camera_IP:80/stream/snapshot.jpg along with a username and a Base64 encoded password.

Once I’d tracked that down VideoVelocity did it’s thing, and I left it alone for 9 hours. It’s limited to low res images, and some of the fancier features are only available for licenced users. But it’s not bad! ⊕ $95 was a bit steep for me. Especially if it turns out I can do it in R

Here’s the result.