Terry Duell wrote:
Hello Paul,
On Wed, 27 May 2015 08:40:18 +1000, Terry Duell <[email protected]> wrote:
I'll have a look at your project.
I think it all boils down to 'good' and 'less good' control points.
I did a little editing of control points, and removed the vertical and
horizontal CPs, and the optimisation now looks pretty good. I did have to
correct roll using 'move/drag' in the fast preview window. My project file is
attached.
Automatically finding good CPs on maps can be a bit problematic. For some
projects it can end up being less work manually adding the CPs.
I did use the new 'EditCP' tool in hugin-2015.0.0, which made finding and
removing the worst of the CPs an easy task.
How did you do this? Each image on the preview is tiny (since I have 30
images), so I don't
see how you got benefit from the tool.
You clearly know something I don't.
Teach me, oh master!
(in general I suspect people don't use the CP editor much; on my laptop,
with a 1366x768 display, the image windows are only 669x381 pixels
in full screen mode.
(and see my other request for a "don't show the contrast enhanced square in this
window") option.
This result would suggest to me that the mosaic mode is working OK.
Cheers,
>
Adding just one vertical line, in the one image, is all that is needed to fix
it.
Mean error 0.5, max 2.9.
Perfect!
The result is also surprisingly square, given the lack of all but 1 vertical
and horizontal CPs.
Well, except it's upside down...
For purely cosmetic reasons, I gave it a 180 degree rotate (in move/drag),
and centred it (in move/drag via X/Y edit)
Anyway - I tried to center the pano, using mosaic move/drag to alter the X Y.
I just though I'd do an optimise, just to be sure this "finished" pano was
stable.
And after 213 iterations, it converged again, upside down, pretty much back
where it started.
There is "some kind" of optimisation attractor which I don't understand, such
that X Y are interacting with Y P R.
I had thought (till now) that X Y Z only really had meaning realtive
to each other (so that the position of the anchor was entirely arbitrary),
but given the existence of such an attractor, it seemed there must
be a "best position" for the anchor (which implies that X Y coordinates
are not just relative to each other), so I added X and Y for the anchor
to the optimisation variables and let the optimiser loose.
It converged:
avg: 0.008403
std: 0006875
max: 0.0617
with (anchor) X=-85, Y=0.7
which sounds suspiciously similar to the sort of optimisation
that results in a tiny FOV (massive focal length). It just puts all the images
in a tiny dot.
It looks like the mosaic model makes the mapped images smaller (and hence CP
pixel
distances smaller) as X and Y are further from the origin. I believe this is
the driver of
the bogus attractor I'm seeing.
BugBear
--
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/hugin-ptx/55659252.7030705%40papermule.co.uk.
For more options, visit https://groups.google.com/d/optout.
