But for alignment, optimizing for shift is sufficient; the resulting
image transformation is exactly the same as when shifting the crop, only
the masked circle is not moved.
I assume the lens doesn't move between shots, so a global optimization
for shift should be sufficient. PTGui now does this by default.
Joost
> Yes that might be useful.
>
> But for alignment, optimizing for shift is sufficient; the resulting
> image transformation is exactly the same as when shifting the crop, only
> the masked circle is not moved.
>
> I assume the lens doesn't move between shots, so a global optimization
> for shift should be sufficient. PTGui now does this by default.
My lens actually moves, or rather bends/sags, because it is so long and heavy (Raynox).
I typically shoot 4 + zenith(60 degree) in HDR, and the zenith shot(s) have a slightly shifted circle.
I usually start with the same crop circle for all.
Then I go to the zenith shot and just move the crop circle (individual + HDR-linked) slightly to make it fit better.
Works very well, and isn't much work :)
It does work pretty well without that special crop/shift treatment as well, but the optimal is to use a different shift.
--
Bjᅵrn K Nilssen - b...@bknilssen.no - 3D and panoramas
Recognizing the crop circle is easier for a human than for a computer.
An automatic cropping algorithm would not be very accurate: it would be
disturbed by black areas near the edge, or by lens flare reflecting
inside the lens. So the automatic adjustment of the crop circle would
probably cause greater errors than the actual movement of your lens.
Joost