Image stabilization for video is different than image stabilization for stills. That’s because the nature of the shake to be corrected is different. For video, the correction has to last longer and the shakiness may very well be more intense (like when you’re walking and shooting video at the same time, for example.)
That’s why most modern digital cameras have a different, more aggressive system for correcting video shake. In addition to whatever optical stabilization the camera offers, the more aggressive way takes advantage of the fact that the number of pixels in a video (even a 4K video) is far smaller than the number of pixels the sensor has. So the camera might shoot a little wider than what you see in the viewfinder, and the camera compares adjacent video frames, aligns them (much like it does in HDR mode), and crops off the rest before merging it into the video stream. All in real time. The more you oversample (capture pixels outside of the intended frame), the more leeway you have for fixing a shaky hand.
I was reminded of this when I shot a very informal video giving a tour of my studio. Here it is, shot using an RX100 V and a small tripod-socket-based handle (mechanically similar to the VCT-SGR1 recently announced by Sony with their RX100 III vlogging bundle,) For this, the more intense "Intelligent Active" SteadyShot feature for movies was enabled:
As you can see, the shakiness of my hand often exceeded what the camera’s made-for-movies steadyshot could do. It bugged me a lot, but when I posted it to my youtube channel, only one person complained. (On the other hand, this is Youtube, where even the worst production values raises the average.)
Is there anything I can do in post-processing to smooth this out?
