I agree with Errvald. Back when Myst was released due to the lack of power back then, they had a very simple setup that worked great and was found in tons of games thereafter. In Myst, they had 4 different hit boxes on the edges of the screen and had an internal hierarchy of which rendered image would lead to the next.
Now if you want to replicate that, simply render each static shot, and replicate the same set of hit boxes on the edges of the screen and simply render it to the screen with the HUD’s “Draw Simple Material”.
Otherwise, your other option to do it in 3D but to simulate the same type of movement limitations (ie, having to click edges of the screen to progress, not manual camera movement), you would use the HUD class to register some hit boxes and wire some events to see if the mouse falls within either the top, left, right, or bottom of the screen and if so, proceed in that direction with an associated camera animation if they click within those areas. You would also want to change the mouse cursor to indicate that the user can take action by clicking the mouse.
One last thing, if you go the route of static images and videos, then create your custom UClass that can act as a container for each “location” or scene in C++ (if possible) and create an array of pointers for each possible direction and point the memory address for each to the corresponding one. Kind of like a linked list but more connected children instead of a single one. Or simply store a byte index that references an array that holds all of them for the current level and reference each that way. Hope that helps buddy and good luck.