Re: Computer Compositing

[T3D Larry Berlin <lberlin@xxxxxxxxx>]

john bercovitz  writes:
>I'm a little concerned about moving an object, which has its own
>parallax associated with it, to a different depth.  An object's 
>own associated or shall we say "internal" parallax is a function 
>of its distance from the observer so I think if you change the 
>distance of the object, some type of distortion must result.
>John B

As long as we're talking small adjustments there shouldn't be
noticable shifts or distortions. The prime distortion is that object
roundness and texture may be made perceptible where in the original image it
may be far less available. It's more like a correction to some of the
distortion originating in the camera/lens/film process. Obviously one needs
to do these things within an existing range or scale of parallax or it
becomes an abstraction instead of representation.  However abstraction is an
artform when used as such deliberately.

In terms of in-the-moment interpretation, the mind applies its own memory
of parallax and structure to distant objects and in effect enhances or
amplifies, without distortion, the visible parallax of these distant
objects. In order to duplicate this artificially in an image you have to
place more available cues where the eyes can access the information, in the
image. To a certain extent, depth placement will override any perceived
distortions from small exagerations of surface parallax. If these
exagerations are from actual photo sources it's naturalness will help it to
be seen as normal unless it's too far out of scale with the rest of the
image. Roundness in a distant tree is easy to accept unless it's way beyond
round. The goal is to accomplish a suggestion of roundness as opposed to
obvious relative flatness in something you know to be round or a measured
enhancement of shape where the image capturing process diminishes such
information. (less needful perhaps in slides and projection under ideal
presentation conditions)

We can look at a distant tree and with a moment's concentration access more
depth cues than is recordable in an ordinary photo. This means that *normal*
parallax to the mind is relative to remembered shape as well as directly
perceived shape and an assembled shape from various minute cues including
motion over brief time intervals. This extra resolution can't be added
directly into a photo. Unless one can play parallax tricks with portions of
an image by enhancing distant object shape-parallax as if you were looking
at the original scene and *studied* the distance for the additional cues
that build a more detailed mental image. If you don't enhance the parallax
of distant objects, the relative flatness with distance can be far more
apparent as *flatness* in a photo or image than it ever is to the eyes+mind
while observing a real scene.

Larry Berlin


------------------------------