Experimental logo removal.

[goldenpi]

http://birds-are-nice.me/programming/voronoi.shtml

It's still in the prototype stage, but the mathematics behind this is completly new - never before tried. In theory, it could be a very capable filter for removing logos from animation. If you use firefox or another browser that'll handle animated PNGs, you can see a little bit of video showing what it's capable of. Surprisingly effective!

Thoughts, anyone?

[Superb]

Looks similar to Photoshop's content-aware fill. I wouldn't call it "never before tried"...

[goldenpi]

It's nothing like the content-aware fill, and it is very much never before tried - I searched extensively. Though it may be that there is good reason it's never before tried: There are less processor-intensive ways to achieve as-good or better results. This is really more of a proof of concept. But it does work, and surprisingly well, so concept proven.

[Guest]

Seems to me that your searching skills need polishing:

http://www.google.com/search?q=voronai+inpainting&rls=com.microsoft:en-us&ie=UTF-8&oe=UTF-8&startIndex=&startPage=1#hl=en&rls=com.microsoft:en-us&sa=X&ei=O-HPTpi6B8He2QXe2PjADw&ved=0CBgQBSgA&q=voronoi+inpainting&spell=1&bav=on.2,or.r_gc.r_pw.,cf.osb&fp=fcbd9245a103963f&biw=1334&bih=766

[goldenpi]

Check harder. Lots of uses for voronoi diagrams there, but none for use in disocclusion. I am the first to try this. That doesn't mean it's actually a *good* idea - as my little test shows, the results at at best comparable to existing techniques and yet take an impractical amount of processor time. So this isn't really a *good* method. But I am the first to try something new, even if it didn't turn out to be of much practical use.