I would only add a second stabilization stage for a specific residual problem. In-camera electronic stabilization has already analyzed motion and usually cropped the frame. A second stabilizer does not start from the original sensor view; it works on the already-corrected result.

That can be useful for a gentle cleanup, but aggressive settings can amplify warping, crop and delayed-looking camera motion.

Check the shutter behavior too. Electronic stabilization works better when individual frames are reasonably sharp. If the source has heavy motion blur, post stabilization may make the camera path smoother while the blurred detail still looks unstable.

That is not something a stronger stabilization percentage can truly fix.

In VSDC I would compare three short versions: source as recorded, light post stabilization, and stronger post stabilization. Look at edge crop, geometric distortion and how natural intentional turns feel.

If the light pass only removes a few high-frequency bumps without changing the overall camera motion, it can be useful. If it starts fighting deliberate movement, stop there.

That is exactly what I see. A light pass helps the vibration, but the stronger setting makes every head turn feel delayed.

I will keep RockSteady as the main stabilization and use post only as a selective cleanup on the worst sections.

Selective is the key word. Stabilization is not automatically better just because two stages are available.

Preserve intentional camera movement and spend the second pass only where the first stage left a specific defect.