It's a fascinating idea to transpose audio granulation into video.
If we think about it this way, there are not many differences between the two media.
Audio is a temporal phenomenon, and while a single frame is fixed in time, a video can be considered a temporal phenomenon as well.
Parameters typical of audio granulation transpose well to video:
• amplitude becomes brightness
• frequency shift becomes playback rate
• Freezing becomes holding a small set of frames
and so on.
Granulating sound lets you inspect the smaller parts that compose an auditory experience, a bit like separating the sound of each leaf in a tree rustled by a gust of wind.
You can then play all these little sounds at once and overlap them to create something different from the original material.
With video we can visualise the single moments that compose what has just occurred, and then play them separately or all at once to re-imagine what has just happened.

It is something that doesn't seem to have been widely explored so far, and I am having a lot of fun finding technical solutions and being surprised by the output of this idea.

Federico