Basic functionalities:

- get the js code to work with DRF (full-AJAX)
  → WIP, hackish but works somewhat, coordinate reference is wrong (pixel in backend, x in 0..1 and y in -0.5..0.5 in frontend)

- hire a HTML5/CSS designer :-)


Bugfix:

- when the image does not loop, it should be displayed only once in the
  interface (right now, it loops, which is ugly and impractical)


Background work:

- use Celery for computation-intensive tasks (generating tiles)
  => in progress.  We should display the progress of tiles generation
  in the panorama view.

- general js code cleanup (tile display, bearing/elevation computation)


New features:

- Add photos, URL, etc to ReferencePoint model (useful to collect
  information about the reference point: pictures of the building,
  webpage with description, etc)

- Mini-map on the side of the panorama view, many uses:
  - see the current direction of the panorama on the map
  - click on a point on the map to see this point in the panorama

- Create new reference points directly in the panorama view (right click /
  new ref point / ask coordinates, name, etc).  Even better: open a pop-up
  or a new page with a map, so that the reference point can be placed on it.
  We could even fetch the ground altitude automatically, and only ask the
  altitude above ground.

- Allow to zoom more on tiles, even if this means having a very low quality

- Given a new point (for instance, a new person that wants to be connected),
  list all nearby panoramas that point in this direction, with links that
  orient the view automatically towards the target point

- ability to click and drag a reference point in the UI when it is not
  at the right place