Hi, Martin! Thanks so much for creating this wonderful program. I've already used it to create a series of 'portraits' I'm very happy with:
I have a similar question to chelioz from earlier, as I've been wrestling with the question of 'why is it that analog slitscan processes are so much easier to get high-speed results with than digital?' Since film doesn't have a 'framerate' to deal with, just the speed at which you pull the film across the sensor, that's obviously a big reason. But I also realized that analog slit-scan cameras don't have to deal with the problems caused by 'slit perfection,' as your 1-pixel-at-a-time algorithm creates. Analog slits are never going to be small enough to hit only one grain of emulsion per unit motion.
So, my question would be, as a non-programmer—is there a way to 'widen' our digital slit at all? In my imagination, it would involve first pulling more than 1 pixel per frame (as many as 10-30 I imagine would still represent a pretty narrow slit given a 4K vertical dimension) and then overlaying some percentage of those pixels (perhaps with an 'addition' blending mode—this would require underexposing the original video to wind up with a correct exposure) frame by frame.
Would love to hear your thoughts.