iso rivolta
Senior Member
I've got the same idea some time ago. More specific, clocking the overflow drains at individual pixel level. Of course, the circuitry must be more complex, but these type of gates already exist on many sensors, implemented mainly on CCDs for avoiding blooming artifacts. So, not like the CID sensors or the Samsung patent, where, even with nondestructive readings, dead times still exist. And for other types of sensors, the more readings method would introduce more noise. On the other hand, that would be nice - imagine live "light painting".
One possible mode of functioning for this proposed type of reading would be by storing the digital time values of reaching the full well at pixel level in a memory layer under the sensor. When all the pixels have reached that, the time value array is read. Of course, that would be for maximum quality, for tripod work and still scenes. But the sensor could also function in the normal way or in a more complex hybrid mode - after imposing a shutter speed, some photosites reach saturation, some not. Things are much more complicated, the threshold charge in the drain area is not just 1 electron and there will still be some "read" noise in the time values.
Some simplified tutorials on CCD blooming protection can be seen here:
http://www.olympusmicro.com/primer/digitalimaging/concepts/blooming.html
http://www.olympusmicro.com/primer/java/digitalimaging/ccd/blooming/index.html
One possible mode of functioning for this proposed type of reading would be by storing the digital time values of reaching the full well at pixel level in a memory layer under the sensor. When all the pixels have reached that, the time value array is read. Of course, that would be for maximum quality, for tripod work and still scenes. But the sensor could also function in the normal way or in a more complex hybrid mode - after imposing a shutter speed, some photosites reach saturation, some not. Things are much more complicated, the threshold charge in the drain area is not just 1 electron and there will still be some "read" noise in the time values.
Some simplified tutorials on CCD blooming protection can be seen here:
http://www.olympusmicro.com/primer/digitalimaging/concepts/blooming.html
http://www.olympusmicro.com/primer/java/digitalimaging/ccd/blooming/index.html