Can’t seem to get something to synth right? Curious about what lens is best? Ask your fellow Photosynthers here.
I just came across this a couple days ago and think it's awesome! but i'm trying to find a program where it will be able to complete a picture by drawing on other pics. kind of along photosynth, but one step further - ex. say you have a pic of the statue of liberty but the torch is cut off at the top and the software examines other pics of the arm/torch but instead of matching up the separate photes like this does, it creates the arm and torch, surrounding sky, etc on the original pic......... is there anything like that out there?? or did I just see it in a movie? thanks. PK
I'm not yet aware of any publicly available programs that do what you're talking about, but the 'Content Aware' editing in the recent Adobe Photoshop CS5 does a pretty amazing job of using other parts of just a single photo. ( see http://vimeo.com/7170672 and http://vimeo.com/5024379 )
I'm sure that if it could take a Photosynth and load it, it could do even better by using the information from the other photos to fill in holes or extend the edge of a photo like you're asking. I had similar thoughts a couple of years back, thinking about filling in over-exposed, under-exposed, or low resolution portions of one image with details from others. I was thinking of some earlier research that I had seen where the same 'take detail from other regions of the same image' technique had been applied. ( http://getsatisfaction.com/livelabs/topics/what_mashups_do_you_hope_people_build_using_photosynth )
Since then, I've seen two other very interesting research projects that have begun to show what applying these concepts to video looks like.
First, 'Unwrap Mosaics' from Microsoft Research (see the video at the bottom of the page):
Second (and a personal favorite), 'Using Photographs to Enhance Videos of a Static Scene' by Pravin Bhat who was studying at University of Washington and now works at Weta Digital
Lastly, 'Interactive Video Object Manipulation' from Adobe researcher, Dan Goldman
Once these things are combined with 'Lucky Imaging' (as seen in this video from Lowry Digital [ http://www.lowrydigital.com/enhancement.html ] and from 8:50 to 10:45 in this video from Microsoft Research [ http://ch9.ms/BFV8 ]) as well as motion blur cancellation (as seen from 6:30 to 8:00 in that same Microsoft Research video) you can effectively expand the field of view and increase the resolution and|or clarity of a video by taking the detail from the sharpest and highest resolution representation of anything in the video and flowing that detail back into the places where it is farther from the camera and I assume that the same applies to a good robust synth where you have a good clear point cloud of the objects that you're interested in.
I'm not sure how the current version decides which patches to map back into the model, but Yasutaka Furukawa's work at University of Washington on generating dense reconstructions from a sparse structure-from-motion model (like the ones that Bundler and Photosynth generate) seems to me like the steps toward the future of this whole field where the best quality representation of anything can be mapped into the same model and retained.
(watch the video for 'Towards Internet-scale Multi-view Stereo': http://www.cs.washington.edu/homes/furukawa/ )
Anyhow... I know I wandered around a bit, but these little scraps of research are what came to my mind when I saw your topic. I hope you find them as interesting as I did.
If anyone knows of something that is available today that does what P_Kieda was asking about, please chime in and let us know.
i think i see something about it on http://www.aexlimo.com or on youtube i do not remember yet but i will find it
http://www.elitechicagolimo.com best Chicago Limo Service for any occasion. Check it out for yourself and you will surely love our limo service in Chicago.