Am 03.01.2011 00:17, schrieb syntotic:
> So! For a bmp image 595x345,
> placed as a pigment image_map, interpolation 2, once,
> translated [-.5,-.5,0] and _scaled_ 1330*<595,345,1>/345,
> on a plane {z,2300}
> seen through a camera at [0,0,0] looking_at [0,0,2300],
> the original image displays well at size 597x345
> with top corner located at 101,127
> on a render size image of 800x600.
>
> You get the idea... Height was adjusted, but width does not fit without losing
> the vertical adjustment. Translate the image_map [-.6677,-.1367,0]: now it is
> aligned to the top corner for a 800x600 render image size, but dims become
> 598x346. See the magic numbers? The script is brittle though may be enough for
> my particular purpose... One pixel error in size does not seem much, but given
> all other there possible adjustments... Once upon a time I had an image cropper
> utility for these cases to ignore the black background and extract the image
> only, (and a digested C++ COM controlled POV RAY version) but not anymore.
> There _must_ be a simpler expression to locate a picture in a scene! I suspect
> such expression may be quite simple for camera types other than default.

Have you ever had a look at screen.inc? If I understand your problem 
right, that might be what you want.

Post a reply to this message