Cool research on using Photographs to enhance Video

Luis, thanks for sharing! I wonder how long it took to make these examples. I'm guessing it isn't as simple as dropping in the video, photos, and analyzing them with an easy to use piece of software... but it is extremely promising technology.
 
wow, that was excellent,

from the video, it seemed:

they motion tracked the orginal footage (that can be done using PFtract, syntheyes etc) PLUS extracted the Z-depth automatically (possibly this could be done in PFtrack maybe using optical flow, not sure)

THEN they took the still photos and then used (effectively what I would say is Projection mapping) back on the new moving zmap and then finally blended the original video image to remove any missing patches.

However it seems they managed to automatically figure out the camera pose for each still (that would be hard to do manaualy, aka eyeballing)

Good stuff, this is the kind of stuff that eventually does lead to commerical applications.
 
You must log in or register to reply here.
Back
Top