When the Cassini mission flew by the planet Jupiter in late 2000, a sequence of full disk images were taken of the planet. Assembled with proper spatial and temporal registration, the sequence could produce fourteen distinct images suitable for wrapping around a sphere.
But the time steps between images were large and exhibited significant jumping. The solution was to create additional images between the existing set by interpolation. But simple interpolation would not work due to significant changes between the images.
To solve this, we interpolated between the images using the velocity vector field of the cloud images. The velocity vector field was computed by performing a 2-dimensional cross-correlation (Wikipedia: Cross-correlation) between the images. This velocity field was checked against Jupiter velocity profiles from the scientific literature and agreement was excellent. With the addition of a simple vortex flow at the location of the Great Red Spot, the interpolation process was used to generate intermediate images, increasing the total number of images from 14 to 220 and resulting in a smoother animation.