On Fri, Oct 03, 2014 at 01:08:33AM -0700, T. Modes wrote:
> * Optimizing TrZ *and* HFOV for all images with separate lenses is not
> helpful. Changing TrZ and HFOV have both a very similar effect on the
> remapped image (both scale the image). So the optimizer can achieve the
> same result with 2 different values. This makes the optimisation more
> fragile than needed.
"more fragile" is sort of an understatement. When the size of an image
ends up like size = C * TrZ * HFOV, optimizing for both should
converge on any solution where TrZ * HFOV comes out to the same value.
Then the optimization is in a difficult situation. Each time it has a
tentative solution it will figure out if the solution becomes better
or worse when increasing or decreasing each variable. Well, the
solution will have ALMOST no change when decreasing HFOV as long as
you increase TrZ proportionally. But going the other way also has a
similar effect. In practise the computer will always find a minimal
effect, and end up with one of the variables huge and the other very
small. Small computational rounding issues will determine what way
things will go. But you'll usually end up with a "silly" solution,
that exagerates any computational roudings. So you'll end up with
something like: "Oh, you took this picture from the moon, with a very,
very good telescope!", but then tilting the telescope just a faction
of a degree will make the placement of the picture totally wrong.
That sort of stuff.
Even more practical, the transformations are probably not identical,
but only very close. So slight mis-positioning of a control point
(even at fractions of a pixel) will cause the algorithm to go haywire
into some silly direction.
Roger.
--
+-- Rogier Wolff -- www.harddisk-recovery.nl -- 0800 220 20 20 --
- Datarecovery Services Nederland B.V. Delft. KVK: 30160549 -
| Files foetsie, bestanden kwijt, alle data weg?!
| Blijf kalm en neem contact op met Harddisk-recovery.nl!
--
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/20141003093614.GD23689%40bitwizard.nl.
For more options, visit https://groups.google.com/d/optout.
