How does 2D to 3D conversion work?

[TomFahey]

So, I was wondering how you go about converting 2D content to 3D - as in, how does it actually work? I read about how, when they convert an existing big blockbuster film, they manually convert each of the frames into greyscale depth maps, but I was thinking more along the lines of how the real-time conversion methods work, as in the ones you get in most 3D tvs and projectors nowadays? I thought that an interesting approach might involve applying the same sort of analytical system that the auto-focus in dSLR cameras uses, by measuring the amount of contrast between adjacent pixels and working out how far out of focus they are, and therefore how much "further back" a particular object is from the object in focus. Obviously this would require the object that's in focus to be a known distance away - I was just wondering if the real-time 2D to 3D converters work along a similar principle?

[android78]

The short answer is - It doesn't. At least, that's my experience on a samsung plasma. It seems to make flat planes on the bottom half of the screen have the bottom closer to you then the middle of the screen, and the inverse for the top half. For anything that doesn't appear to be a flat surface, against the flat surfaces, it seems to make them flat-ish with the depth being dependent on the depth when it touches the flat surface.
It also seems to do some adjustments with shadow estimation, but they seem pretty dodgy to me.
It has some reasonable effect for things like sports, but is pretty much a waste of time and not worth it IMO.

[xhonzi]

I know my Monoprice converter uses some form of the Green Lantern energy to do its conversion.

Because it's really terrible at converting anything that is yellow. Other than that, it seems to work pretty well. But seriously, the yellow stuff is always at maximum pop-out.