> Hi Guys,
> 
> Was doing some experimenting with the vshift and hshift parameters on ptgui 
> and I noticed a vshear & hshear parameter...... What are these?? How do I 
> find them for my cam?
> 
> Sam

Sam,

This is what Helmut wrote some time ago:

"Image shear (which is one of the options of the PanoTools plug-in
'Correct') can now be optimized and applied to the Stitcher. [...]
This is useful for slide and transparency scanners. These inevitably
exhibit some slight misalignment of the line scanner relative to the
film transport, which results in this type of distortion, and which is
not covered by any of the other correction methods. Not visible in the
original scan, but may easily add a few pixels misaligment in large
panoramas."

I don't think it is useful for digital cameras.

Below is a message I posted to the imim-list a few months ago, may be
interesting.

--------------------------------------------------------------
I've tried to determine the shear parameter for my film scanner (HP
S20). I think it's safe to assume that the image shear is constant for
the device. Especially when scanning film strips, since the film strip
can only move in and out the scanner, it's almost impossible to rotate
or slide it sideways. So the angle between the movement direction of
the film strip and the CCD is more or less constant.

I made two scans from the same negative, the second scan with the
negative upside down, so you get a mirror image. This second image was
mirrored back in the scanning software. This results in two similar
images, but with opposite shear errors.

I set about 20 control points for the images, and used the following
script:

# script file for ptOptimizer created by ptGui 
p w2000 h1400 f0 v70
u20 n"PSD_nomask" 
i n"Scan154.jpg" w3400 h2248 f0 y0 r0 p0 v64 a0 b0 c0 d0 e0 g0 t0 
i n"Scan158.jpg" w3400 h2232 f0 y0 r0 p0 v64 a0 b0 c0 d0 e0 g0 t0 
(control points here) 
# optimizer: 
v  v1 d1 e1 g1 y1 r1 p1

As you see I optimized the field of view, shift and yaw/roll/pitch of
one image, to correct for differences in the (manual) cropping of the
scans.

Further I optimized the horizontal shear of the second image. I found
out that you should not optimize horizontal and vertical shear
simultaneously, since these more or less compensate for each other and
the optimizer will start drifting. I figure that a misaligned CCD will
introduce only horizontal shear if the image is in landscape
orientation, or only vertical shear in portrait orientation.

Optimizer result (second image):

o f0 r-0.038088 p0.0560778 y0.0314704 v64.0051 a0.000000 b0.000000
  c0.000000 g8.709629 t0.000000 d-11.621583 e26.863518 u20 +buf 

Control point distance is less than 1 pixel for all 20 control points,
so there's a perfect match.

So the shear error between the two scans is 8.7, which means that the
scanner has introduced a shear of 4.35 which would mean a 0.2 degree
misalignment of the CCD. 
--------------------------------------------------------------