I would add a proviso to all of this... there are different kind of noise and different kinds of distortion caused by filters, and the appropriate method to bring these out varies with the kind of phenomenon in question. I would say that whatever else you try you should use a least one method which involves playing the clip. (E.g. because strong spatial filters might yield frames which individually looked very good but which do not match the neighbouring frames closely enough to give a fluid result.)
I generally use two playback methods: one method which applies a filter to the right-hand side only (similar to the method in the second post), and one method which switches the filtering on and off every few seconds. I would personally avoid simple stacking methods because the requisite subsequent resizing will make it hard to notice a lot of phenomena. Subtract is useful in learning where distortion might occur so that you can look for it in the filtered clip, but once you have an idea of that I would switch it off, if no other reason than because you can't look at the subtract clip and the original/filtered clips at the same time.
|