This week I made a paper prototype setup for how my popup previews would work. I made representations for "setVehicle" and "resize". In order to do so, I looked through the current posted worlds to see which ones would showcase those actions. In doing so, I'm more convinced that in order for this to work, we would need to be able to automate searching for appropriate clips, since doing so by hand is inefficient and likely few would be motivated to do so. However, I was able to refine what I would look for in defining a "good" clip- I specifically searched for videos where the desired action could be easily separated, ideally in its own Action Ordering Box, which contained only actions for one specific character.
Questions I will be thinking about in the future are:
Does the connection weaken if the character performing the action is different from the user's? (ie, say the user has a bunny, will the user make a stronger connection with a clip of a bunny or a clip of a Cheshire cat? What about an alien? A bowling pin?)
How common is it to have actions neatly separated like this? Is it feasible to have a preview clip (or multiples) for ever character's action?
Would it be possible/useful to select or prefer a particular clip based on the user's current scene? (ie, the user has 3 identical characters, should we favor showing them a clip involving 3 characters doing that action or the "cleaner" clip of just one character doing it?)