Building Rome on a Cloudless Day (ECCV 2010)
Loading...
67,148
Loading...
Uploader Comments (JanUNC)
see all
All Comments (25)
-
When any software will be available ?
and how do you dowloaded (maybe?) all those image ?
-
i knew this could be done ;))
-
It's Colosseum.
-
This + Kinect
-
Needs a shitty thumping techno soundtrack.
-
both this and photosynth completely blow my mind. makes me want to go into CS!
-
@37coolidge For an interesting video about building 3D models of faces based on individual photos, check out the video at watch?v=nice6NYb_WA (I get an error when I try to post the full URL)
-
Thumbs up to your whole team
would like to know about your project and team
Do you have a dedicated blog???
Loading...
Outstanding work by an outstanding team. Have you tried doing this with people?
37coolidge 1 year ago
@37coolidge We haven't since people move hence there is no static object which is one of the underlying assumptions for this. We though in a different line of research work on reconstructing people.
JanUNC 1 year ago
this is pretty impressive. i have two questions:
1: have you tried to exploit shadowing effects from flashlights of cameras to help 3-d rendering from all images. i know this is something archeologists use to help reading faint texts written in stone.
2: how is your technology different from Microsoft photosynth?
broccolee 1 year ago 3
@broccolee
To 1) This can indeed be used but in order to do so in an uncontrolled photo collection you would need to robustly and automatically identify it which is hard to do reliable in practice. Hence we didn't use it.
To 2) The major differences are that we create a dense 3D model and in contrast PhotoSynth uses the images for visualization. Additionally we can handle orders of magnitude more data millions of images compared to thousands. We can handle more disturbances in the data.
JanUNC 1 year ago 4
Amazing job, how long did it take you from start to completion to create this application?
thejacenxpress 1 year ago
@thejacenxpress
Its several years of research in the individual modules coming together. Combining the methods still took 3 month of intensive development but it has been a fun and rewarding experience given my great team.
JanUNC 1 year ago 2