Thursday 13 June 2013, 12:00
My name is Alia Sheikh and I recently blogged about my role on BBC Research and Development (R&D) Production Labs, but I have also been working on the BBC Two programme Horizon: The Secret Life of the Cat.
In BBC R&D we're used to working on cutting edge technology for broadcast use so my colleague Tony Richardson and I were really excited when Horizon approached us to develop the cat-cams which were to be used to track the lives of several cat subjects.Rosie the cat with her cat-cam
They needed cameras that would be light enough to be worn by cats day and night to give the audience a cat’s eye-view without constricting their movement.
Working with the Royal Vetinary College we decided that although the camera weight would ultimately depend on the size of the cat, we should aim for approximately 35g per camera.
We also needed the ability to record a Global Positioning System (GPS) log of the cat’s movements and also make sure we reserved the battery of the camera when the cat was inactive.
Since the safety of the cats was a priority the cameras needed to be on quick-release collars that would easily detach if caught in anything and we also needed some sort of homing beacon to let us find the cameras again if they did fall off.
Initially we considered using GoPro cameras but even the smallest and lightest version is 73g, over twice the target weight. Existing pet-cam products are available but typically produce a lower quality picture than we were aiming for. We also looked at the tiny cameras used in spy watches before dismissing them as too tricky to work with in the timeframe available, we wanted a camera we could easily modify for our own needs.
We therefore chose HD808 keyring cameras which are popular with model aircraft enthusiasts due to their small profile, extremely light weight and the ability to swap out the lenses.
We took a HD808 camera, stripped back the casing to remove some weight and moved the lens assembly from the side-edge of the camera to the front-centre and used 120 degree wide angle lenses, meaning we could get a stable picture on a cat.
We swapped the battery for one that would last longer and removed the infrared cut-off filter (IR filter) then encased the prototype in lightweight foam to create our first prototype.
We lent the camera to the show's producer who fitted it onto her cat and let it roam free.
The initial tests were successful and the camera hung at just the right height to capture the cat's whiskers as it moved about. We later learnt from Dr. Ellis of Lincoln University that the movement of cats whiskers tells us a lot about its mood and behaviour, so it was very useful to have them in shot.Kato with cat-cam
Although our prototype cat-cam was incredibly lightweight it was also quite bulky due to the liberal amount of duct-tape used in its construction. In some ways this was a good thing as the camera was cushioned against the cat’s neck and therefore gave nice stable pictures.
However, we were worried that a bulky camera would get snagged in something or affect behavior, so clearly the padding would have to go.
We also found that the camera battery was not lasting as long as we wanted and after some testing we realised that we could drop down from four light-emitting diode's (LEDs) to one without losing any quality in the image, but saving massively on battery life.
However, we still didn't have a camera that would record all night and the concern was that we would end up with hours of sleeping-cat footage and miss the real action.
The solution was to collaborate with design engineer John Lowe who was developing the GPS units to track the cats. This unit was wired into our cat-cam and meant we could see to turn on the camera and LEDs only when the cat was active. This saved on battery life allowing us to have footage throughout an eight to 10 hour period before recharging.
The GPS units needed to be mounted on the collar at the back of the cat's head in order to get a good signal and our cat-cam hanging under the cat's chin provided a good counterweight to keep it in position. We also added a small radio beacon to the camera so we could find them if they dropped off.
Each successive cat-cam prototype was smaller and lighter than its predecessor, progressing from foam-encased to moleskin-fabric-wrapped to finally small vacuum-formed boxes that sat discreetly under the cats chin.
The cameras were configured to the highest possible bit rate settings and a series of tests determined the best settings for daylight and night time filming to get the highest possible picture quality.
The cameras had to be taken in after each night time or daytime filming slot so that they could be recharged, so a certain amount of time was spent waiting for the cats to come home.
For daylight filming we needed to put back the IR filter which we had removed from the back of the lens. The filter acts to block out infrared light which can wash out an image taken in daylight. After a number of unsuccessful experiments making slip-on IR filters for the front of the lenses (all of which were lost by the cats) we decided that it was simplest just to swap over lenses.
Lenses with IR filters stuck to the back were used for daylight filming and lenses with the filters removed used at night. They were refocused with a pair of tweezers each time we swapped them and the camera was hooked up to a computer so we could flip between the day and night settings.
At the end of a day’s filming we’d decant the footage and John would copy the GPS data across. Using this data, researchers from the University of Lincoln were able to quickly determine the cat’s movement and behavior in different locations which was taken into account when John reprogrammed the collars for the next round of filming.
Looking back, having an easily accessible test-cat would have sped up the development a lot. We were delighted with the footage we got back from the cats, one thing I have learned from this project is that a cat actually makes a pretty decent camera operator.
Alia Sheikh is a technologist for BBC Research and Development.
Join the discussion...
Thursday 13 June 2013, 07:00
Friday 14 June 2013, 07:01