Questions about possible functionality of CMOS camera chip

I have no idea if it's the right forum, but this is a subject impossible to find answers for.

I had an idea for app development which requires acquiring information using the phone's camera. Smartphone cameras use CMOS sensors, which (usually) work with an electronic rolling shutter, that is, the pixels are exposed to light and then read and reset line-by-line, going from one end of the sensor to the other.

This leads to a delay in moment of light capture between the first line of pixels and the last. I wanted to make use of that effect by recording several points of time in one picture across several tenths of a second. The issue is that the "shutter travel" happens very quickly, too quickly for what I need; I don't know the exact number, but it's at least as quick as 1/90,000 of a second on my Mi 6 as I can get an image exposed for this little. My question is whether anyone familiar with development of camera-using applications knows if you can control the speed of that operation. It's not like performing something more slowly should be an issue in-and-of-itself, but someone elsewhere claimed it might be hard-coded by the camera manufacturer (Sony for example) to perform as quickly as determined possible. I need the electronic shutter to be slow moving and one pixel-line wide (which basically involves setting the exposure time to the shutter travel time divided by the number of scanned lines of pixels).

Now, if that's impossible to encode, my other idea is to record very high FPS (several thousands a second) at very low resolution, perhaps by reading only the middle portion of the sensor. Even 10x10 pixels might be enough. The question is also whether that is possible? A simple observance of the fact 2MP can be recorded today by most smartphones at (at least) 30fps shows that the bandwidth at least is well within the limit with 100pixels at 10,000fps. Also the shutter of course scans the pixels much more quickly if it can open about 3,000 lines of pixels in >1/90,000 of a second, but I don't know whether the process of saving a pixel takes a determined minimal amount of time regardless of how many pixels are handled in parallel.

Sorry for the long-winded questions. Thanks in advance.

Show Accepted Answer

Guest Quick Reply (No URL, BBcode or HTML)

Last post by mthompson
4 hours ago
Last post by ernesto.daniel
2 hours ago
Last post by monahan.benny
1 hour ago
Last post by kiehn.kylie
3 hours ago
Last post by effertz.corbin
1 hour ago