Ken wrote in message <36C8CD3B.FF291F00@pacbell.net>...
>I have found that the radial distortion you would expect often times
>manifests itself more often with camera positions close to objects
>and seems to deminish with distance. An example would be a tall object
>nearly off screen and near to the camera while the main focus of the
>scene is further back. You can really exagerate this (to me unwanted)
>effect by adding a little tilt to the camera. Like placing it at y*5
>and have your look_at set to y*1.


I also find it an unwanted effect for my current effort, and upon studying
the original woodcut I found that the back end appears 1/3 the size of the
front end, and a little geometry told me that the camera needed to be half
that depth away from the front face to get that effect.  Then I "zoomed out"
(reduce the camera length vector) to get the whole scene into view.

When planning it, I thought I might have distortions and was contemplating
ways of dealing with it.  But lo and behold, there is =zero= distortion.  So
my understanding of what causes wide-angle distortion is flawed.

When I get unwanted perspective effects, I get the results I expect by
backing away and zooming in.  So some of it anyway agrees with my mental
model and real-world camera experience.

--John

Post a reply to this message