Michi Knows

I just saw a really interesting video showing off “content aware” image resizing.

The idea is that an image can’t be readily resized without distorting it or making it too small to make out. The author of the research suggests instead to use algorithms that detect and remove less “important” parts of the image. For example, background space between two people might get removed, pixel by pixel, as you shrink an image.

The idea has an amazing amount of potential for use in commercial applications, especially mobile devices.

See the video for more.