efficient stitching should take as input the stage coordinates if they are available.
this will predetermine the overall organization and save a lot of work.

=> all stages should be read when a picture is taken and stored as metadata. 


=============

trackEM2



=============

http://pacific.mpi-cbg.de/wiki/index.php/Stitching_2D/3D



==============

xuvtools




==========================================

endrov gui way:
if there is a data explorer window (really needed!):
select all imagesets (channels?) -> right click -> "create stitch"

stitch objects will reference objects and render on-the-fly

this calls for new way of rendering in 2D, using pluggable renderers. this is one step closer to
2d-3d integration. keep one window, two views?

how not to show original images + stitch at the same time? let user disable originals? maybe easiest

one imageset = one position? most likely, multiple imagesets per  