My comments, questions and enhancements

[Peter Suetterlin]

Ed Halley wrote:
> 
> I've noticed this tendency myself.
> 
> If Hugin wanted to avoid this, would it be useful to do the equivalent
> of 100 iterations of guesswork (y,p,r), get some better estimates from
> that, and then try to optimize the fov and other parameters?

Especially I rate it very helpfull not to optimize (y,p,r) together in
the first step.  I often end up with images that, instead of
(-180,0,0) get something like (0,-180,0) (not (0,-180,180), which
would be the same, but without rotating the image - i.e., it's upside
down on the other side).

> In the meantime, I would suggest that Hugin realize that there are
> disastrous results from optimization sometimes, and not even offer the
> user a chance to "Apply" those changes to the current project.  If it
> could record a disaster/singularity flag in the pano or opts, then the
> Druid could try to explain how the user can get better results.

Hm, not sure.  Currently you cannot really see aht is wrong before you
actually 'accept' the results.  And unless you know what's wrong you
have a hard time trying to correct it....

Btw., the Druid complains that my panos are too large and I should use
'calc size' to get an optimal one.  Of course, that would result in an
even larger pano....  Druid should check that himself and compare the
size with the optimim size, not with a fixed one...

  Pit