I would like to extract multiple frames from a video and combine (say via add) them in one image. I need to be able to control the extraction time interval, the number of extracted frames, moments of extraction, etc.
Sounds like you want a sort of feedback chain that you can switch on and off, but you probably need a threshold too. Even when things are totally dark, camera noise will build up in the feedback chain pretty quickly. Matrixset could be a way to play with it as well