You could use two triangles to form a rectangular
section, and use UV-Mapping to stitch the corners
of the image-map to the corners of the triangles.
So, placing the triangles in any fashion will distort
the image to your liking.

-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

> The other day on a science channel I saw a demonstration of the VISAR
> image stabilization and enhancement software/hardware.
> Visar Story

> They used it to look at a licence plate, stabilize a guy's water skiing
video, and
> some other stuff.  They explained that it takes information from multiple
frames to
> form the higher resolution output.  The demonstration of the priciple was
with an
> overhead projector and multiple blurry transparencies of Albert Einstein
that when
> lined up formed a clearer image.
>
> So I figured this sort of thing was right up POV's alley =)  Just map a
few images
> and filter a bit and voila!  Unfortunately, I found that judging off hand
the placement
> in a scene of a flat box to correct for perspective correction
unrewarding, even if the
> image_map seems lined up, when it is squared to viewer (bringing image
along) it is
> obvious that the image is still distorted.
>
> What I would like to do is map the image to an object first, and then
stretch it to be
> correct.  In other word given 4 points representing something rectangular
in RL that
> is in a image, to then stretch the image to match some other set of 4
points (hopefuly
> correcting for perspective in the process)  Is there some way to do this
sort of thing?
> Any input would be helpful.
>
> Tim Attwood
> #macro T(I M)cylinder{0,I*x,.2 rotate 15*M*z translate A pigment{rgb
E}finish
> {ambient 1}}#end#macro Y(E O)#local A=y-x+O;T(1,0)T(2,18)T(1,12)#local
A=x+y
> +O;T(2,17)T(2,19)#local A=.7*x-.5*y+O;T(.6,0)#end#local D=seed(2);#local
W=1;
> #while(W<35)Y(rand(D)*z<9*rand(D)-5,9*rand(D)-5,9>)#local W=W+1;#end
Y(1,4*z)
>

Post a reply to this message