Thank you for implementing 1s exposure. It works for me and it is a big step forward in my project. It is about monitoring the night sky for bright meteors and also for using the Pi as a webcam showing the cloud cover during the night and all lightning. I want to capture the whole night with 1fps and 1 sec exposure time and then combining every 20s of video (20 images) to a single image that contains the brightest pixel of all frames. The result looks similar to a 20s exposure with a photo camera (e.g. headlights of moving cars become long lines). I already do this since many years with v4l2 and also with UVC webcams under Linux, but now I want to try it with the Raspberry camera.
I made a test video ( http://www.youtube.com/watch?v=PzYGguBz4LI
) with the NoIR camera yesterday and the output shows some of the brighter stars on the night sky. The performance at night is similar or even better compared to the best USB night camera I used so far (Logitech C905). I plan to exchange the lens of the Raspberry camera with a faster lens in a later stage of the project and this will result in even brighter images at night.
Next I need access to the raw pixel of the video.
I first tried to convert the h264 video to a jpg image sequence with mplayer and then to read all jpgs. But this is too slow (about 1 frame every 2 seconds) only for creating the image sequence. Maybe it is possible to use the hardware acceleration for decoding the video. I was taking a look at hello_video.c ( https://github.com/raspberrypi/firmware ... ello_video
) but so far I don't know where I can access the image data and what is the format of the data.
The other option would be to analyze the images during the capture process. Performance should not be a problem, because I capture only with 1 fps. I downloaded the RaspiCam API from here ( http://www.uco.es/investiga/grupos/ava/node/40
). It works on my Raspberry with and without OpenCV. The only problem is that the exposure time is limited to 0.33s and the images are much too dark (see attachment for comparison). I definitely need 1s.
If there would be easy access to the raw data in raspivid this would be a very much welcome feature. Or maybe somebody can give me a hint what to change in the source code of the RaspiCam API to have an exposure time of 1s instead of 0.33s. Or there are other projects where somebody has managed 1s exposure time and access to the raw data.