In the example video you’ve linked to, the camera motion is not limited to a specific type of motion such as “Tracked Dolly”. The levels are broken into zones, with different camera movement settings for each zone.
For example, at the beginning of stage 1, the camera is typically a follow camera which tries to maintain a relative offset from the player. As the player moves into different areas, the offset changes, which may cause the camera to (for example) move lower or higher, or closer/further away. A little further into the level, we start to see some zones where the camera is either fixed in place or weighted towards a particular position (sort of like the camera is on a stretchy leash that pulls it towards a particular point).
It looks like the camera motion is smoothly interpolated when the player moves between different zones. For example, if Zone A has a high follow camera and Zone B has a low follow camera, the camera will smoothly move from high to low when the player moves from Zone A to Zone B.
In Unity, you can achieve a similar camera system with Cinemachine Brain. Brain lets you create different “virtual cameras” throughout the scene, each with their own motion settings. Brain can then smoothly blend between these camera settings as you move into different areas of the scene.
