This idea just came up to my mind some days ago. The thinking was straightforward. I’m buiding cameras for my PhD project and I love (although I really have no idea about it) open source. I googled this term and found a blog of it. Also, a news was released last September telling us there are some people who already developed this stuff.
Well, I’m glad to see it. At least it appoves that this is not a stupid idea, and someone (in U of Stanford!) considers it serious. But the ‘Frankencamera’ is not so cool as I expected. A demo was given showing it can do auto-chop-and-paste thing, or “Photoshop on the camera”. I hardly found it impressive. I’m not saying it’s a dull camera but I don’t want to programme a camera just to do stuffs I can do off-line.
(Sorry if the order is messy. I’m trying to organize my thought.) So in my opinion, the camera I want to develop is a kind of study camera. It’s not advanced and it’s not expensive. The specification might be:
CMOS sensor, 1024*768 pixels, 8~10 bits, C-mount, USB/firewire interface, size less than 10*5*5 cm, with the price less than 100 pounds.
The potential users are teachers, students and fans who want to try their ideas (algorithm) before they buy the expensive instruments. The flexible parts is you can programme to change the exposure time, frame rate, regions of interest, binding pixels or not, gain, or even determine which regions to look at on the fly(which is implemented in my project:)). Assume the way you are looking for your girl friend in the crowd. You keep your eyes focus on the special target, and all the rest people are blur to you. You can use the camera that way to save the bandwidth and the storage space.
That’s about for this post. I’ll carry on talking about what I found existed on open source cameras and the feedbacks I got from my friends and my supervisor.
I would like to thank Otis and Todd for their kind help and comments. Glad to hear from US:).