On 27/01/12 12:08, kfj wrote:
On 26 Jan., 19:21, Doug<[email protected]> wrote:
I've a set of 5 hand-held images each at 3 bracketed
exposures, taken with the AEB setting on the camera (to
minimise movement between shots).
The images are set up as 5 stacks with unlinked images; the
control points found with CPfind Multirow/Stacks; alignment
optimised for y,p,r,x,y,z; then stitched with exposure fused
from stacks.
The result is fine towards the edges of the panorama, but in
the centre the stack(s) is/aren't properly fused - both
foreground and distance are out of register.
It looks like your take was done best as you could. One thing makes me
wonder, though: x, y and z aren't what you optimize for in a standard
panorama. Try resetting them to zero, reoptimize and see if your
problem is still there. Do optimize for lens parameters, though, (a-e)
- in case you haven't done so.
Could parallax between adjacent images in the centre of the
picture be causing problems? What could I do about it, if it
were?
Most likely this is the problem. But you might still improve the
outcome by trying to align the stacks better. For that, you'd first
create more CPs. The specialist CP generator for stacks is still
align_image_stack, and if I can't get my stacks right straight away
with cpfind, I do at least give it a shot - it's slow but thorough.
And when using cpfind, it may help to run it with my favourite
options:
--fullscale --sieve2size 5
on top of what else you may be using - this should give you up to 125
CPs per pair, nicely distributed over the images. I assume you have
used the multirow feature so that there are CPs for all images inside
a stack and the stacks are only inked among each other via a single
exposure in each stack (usually the middle exposure) - if all images
are interlinked, this can also lead to problems (especially with
handhelp panoramas) - so your images should be linked something like
(1,2,3) (4,5,6) (7,8,9) (10,11,12) (13,14,15) (1,4,7,10,13)
do doublecheck just in case something went wrong with your CPG
settings.
Kay
Many thanks Kay, there's a marked improvement when I
optimise lens params a-e, leaving x,y,z at zero and use
CPfind multirow/stacks with --fullscale --sieve2size 5. (I
haven't bothered to remove sky points because it hasn't
seemed to make any difference with these images).
Nevertheless the foreground and distance are still somewhat
out of register in the centre of the panorama.
I thought to try assigning separate lenses to each stack and
unlinking the lens parameters, then optimising for all lens
parameters including fov. The result was one step forward,
two steps back - the distance is now fine over the whole
panorama, but the foreground register is worse, even
extending a bit beyond the centre.
What would you suggest next ?
Doug
--
You received this message because you are subscribed to the Google Groups "Hugin and
other free panoramic software" group.
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx
