New Video rendering technique using stills

[gsellis]

I posted this in Apps about a week ago and it got about 25 reads. Definitely was not the right forum.

U Washington network "paper" - Using Photographs to Enhance Videos of a Static Scene

A group of U of W folks along with Adobe, MS, and UCB have created a new video rendering engine that uses information from stills to add information to the video. OMG - Object removal, 3d image stabilization, HDR, Super Sharpness, Image Enhancement. There is a MOV showing it and a PDF on the details.

I had debated on where to put this, but I choose unwisely the first time. 😉 I could see this technique being used in Hollywood too for FX. Do the film/digital and do stills. Artist effect the stills to tweak the film parts and you would not have to do frame by frame work as much.

 

[CycloWizard]

Very cool stuff. I use a lot of image processing via MATLAB, but nothing quite this serious.

I've seen a lot of cool image analysis stuff in the last year, everything from this to automatic generation of 3-D meshes for finite element analysis. I'd love to use all of these things, but they're just not accessible. The papers give you the basic algorithm, but it literally takes years to develop the code and use it for anything. Maybe I'll invest the time, then start a consulting company so I can use all this cool stuff.