Obstacles seem to compound themselves to stop me from accomplishing it.
How to pan the scene to find the target spot at a very high zoom? Before I can set Position: X, Y values in Transform effect strip, I must find the target spot. However, once I set the desired Scale to a high value (150×), there is no obvious way to navigate (pan) the zoomed-in scene (either a Movie strip or an Image strip) in order to find the target spot. Output is cropped to the Render Area and zooming out by mouse fails to reveal neighbourhood outside the Render Area. Even if there is some way to turn off default cropping (please, let me know), GUI in the preview area is not responsive to panning by mouse. Trying to reverse the workflow by centering the scene to the target spot first (Position: X, Y), and setting Scale next, is a futile exercise because scaling doesn’t preserve Position values. (Changing the Scale value moves the target spot out of view despite unchanged Position values.) It’s even harder when the resolution is low because all I can see is a blob of interpolated pixels, so I can’t tell where I am and in which direction I should aim thence. (See screenshot.)
Why does exponential interpolation mode fail during superzoom? By its nature, superzoom is an exponential process. But when the Interpolation Mode (in Graph Editor) for Key Frames based on Scale (in Transform effect strip) is set to anything other than basic modes (for example to Sinusoidal or Exponential rather than default Bezier), the result is so erratic that I must presume it’s a bug in Blender (version 2.74).
How to align zooming-panning Transform interpolation curves of two different scene strips (each starting and ending at different time points) to accomplish coherent blending transition from one to another? (From a low-resolution view to a high-resolution view in this particular case.)
I can’t remember other obstacles at the moment. I may add them later.
Should I split this question into four separate posts?
Edit: 8 November 2016
Stacking Transform filter on top of another Transform filter to solve some of the above issues results in transforming a strip cropped to the Render Area by the first filter instead of the full source material.
No one knows anything. It’s as if no one has ever produced a zoom effect in a video. It’s as if there are no Blender experts among the living and Blender had created itself alone. No one knows any other software that allows to get it done. No one even knows what’s wrong with my question. No one knows who else knows. Zooming in on a spot in a video: rocket science.