Quote Originally Posted by Sheiky View Post
Isn't it possible that Adobe theoretically could recognize depth by analyzing the image for the amount of sharpness and contrast?
In certain contrived circumstances, yes. But not for any (or even most) photos. The reason is that there are too many possible inputs (scenes) for the given output (a 2D image). For example, consider a test where you have two perfectly identical photos, but one was made using a large subject far away and the other was made using a small subject that is closer. The result is perfectly identical, so there is no possible way for Adobe to know which one is distant and which is close.

As another example, consider shooting through a chain-link fence. In the narrow f-number version, the fence obscures many details -- there is no way photoshop can get those back by blurring the fence with simulated thin DoF. If a wide f-number lens was used instead of software, the fence would be blurred into nothing and all the details would be visible on the subject (though overall contrast would be reduced, of course).

Hope that helps explain why it's impossible.