It seems that I have located an error in libpano. When horizontal and
vertical shear are both specified, their inverse does not work.
For example, this is a shear of a point with g=100
(First is original point, then inverse, then forward of the inverse;
both coordinates at beginning and end should match)
Point: 375.500000,250.500000 > Inv: 425.599999, 250.499999 Forward: > 375.500000 250.500000 (426, 250)
this is with t=100
Point: 375.500000,250.500000 > Inv: 375.499999, 300.566666 Forward: > 375.500000 250.500000 (375, 301)
but this is with g=100 and t=100
Point: 375.500000,250.500000 > Inv: 425.599999, 300.566666 Forward: > 365.486667 243.820000 (426, 301)
As you can see, when both parms are given, the inverse is sequence inverse/Forward is wrong..
(all other lens parms are zero, and pitch, roll and yaw are also zero)
As a comparison, here it is with d=100, e=100 (everything else zero)
Point: 375.500000,250.500000 > Inv: 475.499999, 350.499999 Forward: > 375.500000 250.500000 (475, 350)
I wonder what impact this has been having in optimization of g,t parameters.


Daniel M. German
http://turingmachine.org/
http://silvernegative.com/
dmg (at) uvic (dot) ca
replace (at) with @ and (dot) with .
