It would be nice to have a tool built in, likely through the uploading process, that could take a split an image up based on transparency. With the addition of the foreground, I like the idea of adding roofs, but most of the files available are not split into individual assets.
Outside of me manually splitting them up through Photoshop, which is somewhat time consuming, there is no way for roll20 not to recognize the entire image that was uploaded as a singular object. Thus, the tokens are always "below" it.
There are plenty of tools online that help with this, or Photoshop scripts that do this. Basically, you just say split the object based on a value of pixels (manually entered by the user) and then if it runs into that many pixels of transparency, it can split the image up into multiple assets/objects that exist in the foreground.
Another nice thing about this, would be that the "bounding box" of the map size would be possibly able to be calculated by the tool so that the placement of the entire image is more or less static which leads to the roof objects being placed in the right longitude and latitude based on the constraint of the map's dimensions.
