Beyond the Code
or, how to win friends, influence people and make a living by writing open source software. It's not just about the code.
Follow me on Twitter as @jonobennett.
Friday 4 September 2009, 4:41 PM
Stanford scientists create open source camera
The camera has been created as a tool for researching computational photography, where the capabilities of a digital camera are extended beyond taking a single shot and storing it, through the addition of computational power.
The example the Stanford researchers give of a useful function to build in is automatic high dynamic range photography, which involves taking multiple exposures of a scene to capture as much detail as possible. This normally requires taking several shots manually, then post-processing the images on a PC. Instead the whole procedure could be built into the camera.
The hardware specification for Frankencamera is modest: A Texas Instruments "system on a chip" running Linux powers the camera. The image sensor from a Nokia N95 cell phone is combined with standard Canon EOS lenses to capture the images. The developer has complete control over exposure, focusing and flash, and because the code runs on the camera itself there's none of the latency that would occur in a remotely controlled camera.
The researchers had previously tried doing their work using mobile phones with built-in cameras, but found the image quality and processing power too low, and the APIs for the cameras too restrictive. Hence the reason for creating an open source camera that they and other researchers can use to explore the full possibilities of computational photography. The group hopes to apply the results of the research back to mobile phones, and their work is supported by Nokia's Palo Alto research centre.
PreviousComments on this post
Didn't Kodak and pals do this a few years ago with the Digita open architecture? You could write some quite complex apps and upload them into various cameras.
Curiously hard to find much on this on the Web, though.
I certainly remember seeing pictures of Doom running on a Kodak camera in the late 90s. There is a strange dearth of information available, as you say, but from the looks of it Digita wasn't itself open source, but was a closed operating system with a public API that allowed control over the camera and its processor. It's unclear whether control over taking pictures extended to choosing exposure settings, or whether a developer could control the shutter in real time, for instance.
Certainly, given the power of the processors used in Digita OS cameras at the time — 66MHz in Kodak's cameras — you couldn't have done on-camera high dynamic range picture generation. Not quickly, anyway.
Sounds a good idea - I hope it progresses to the stage where the CCD sensor matches what Nikon and Canon have.
I think the makers such as Nikon & Canon have been remiss in terms of only really offering the facilities that a non digital SLR has. I for one mostly keep to the lowest value ISO setting, but like to increase when I need that extra depth of field and don't want to sacrifice shutter speed. Seems they could have made this easier to change without diving into menu's.
Also would like to see a golden section grid, rather than one Nikon offer.
Also seems a waste that I have this DSLR with firmware and they only update it with bug fixes. I am sure people would pay for upgrades, but then I suppose they really want us to buy a whole new camera
