Glen Berry wrote:

> I'm not sure yet if the human eye responds to light intensity values
> in a similar non-linear way, but I think it does. I know that human
> hearing is non-linear in its perception of dynamic range, but I can't
> say for sure about human vision yet.

The human eye responds in a fashion that is non-linear but dependent
upon the response of neighboring photosensitive elements.  If one
photosensitive element (rods, I'm pretty sure) fires, it inhibits its
neighbors from firing.  The result: when you look at something bright,
many of your rods are turned off, and you don't perceive it as being too
bright, and when you look around in darkness, pretty much all the rods
are turned on, but they're probably not seeing much, because it's dark. 
This allows people (and other animals, but I'm not sure how widespread
this is) to move from bright light to darkness or combinations of those
fairly easily.

The textbook example of a scene that is affected by this is a daytime
interior with large windows.  Using conventional photography, it's
pretty much impossible to make the room look the way the human eye
perceives it.  If you adjust for the actual lighting in the room, the
outside looks too bright.  If you adjust for the lighting of the outside
as seen through the window, the inside looks too dark.  The only
workarounds involve lighting tricks: add more light to the interior, or
wait until twilight so the ambient light levels are better balanced.

Simple post-processing doesn't work so well here; one must apply
different processing to different parts of the scene to make it look
right.  There's a paper at SIGGRAPH from last year that discusses this:
"LCIS: A Boundary Hierarchy for Detail-Preserving Contrast Reduction" by
Tumblin and Turk from Georgia Tech.

-Mark Gordon

Post a reply to this message