Ok, thanks, all of that helps a bit more. The FAQ could maybe be a
little more clear, but it's making sense now.
On Jun 23, 2007, at 7:43 PM, Joost Nieuwenhuijse wrote:
> Alan Pinstein wrote:
> > Joost-
> > Thanks for the updated FAQ. It's very useful. The more I play with
> > panos the more the FAQ makes sense as well :) And of course the
> > better my output is getting.
> > So, now that I (think I) understand linked images... I have a
> > I turned on "link images" before doing an align. Thus, according to
> > the FAQ, I would expect the SAME set of control points to be used
> > each of the exposure sets...
> No, it doesn't work like that. The control point generator generates
> control points for all images, regardless of the link state. It would
> not make sense to apply the control points of image 1 to image 2,
> the images are different.
> > Most of the exposures got automatic control points, but the problem
> > is that some of the really under-exposed images are nearly all black
> > except for the blown-out parts and thus PTGui couldn't automatically
> > find control points for the A4-B4 images. This causes PTGui to
> > complain about bad control points.
> It shouldn't complain about missing control points, since the
> images are
> But it may indeed have generated bad control points; this is always
> possible, hdr or not.
> > However, the one that I DIDN'T correct turned out BETTER than the
> > that I did correct. The one I corrected got all "wavy" on me (ie the
> > pictures wandered up and down) thus requiring a tighter crop to
> > remove the zenith and nadir dead spots.
> As always, after editing control points and re-optimizing, you may
> to straighten the panorama again. See 'straighten panorama' in the
> Panorama Editor.
> > So it seems that in reality it is working just like the FAQ says
> > (using the same control points from the 1st "set" of images for all
> > bracketed sets) but still displaying the control points for each
> > individual set... I wonder if in "linked" mode PTGui is still
> > "aligning" all 4 sets? Maybe this is a slight bug that if fixed
> > actually increase performance (by skipping aligning of a bunch of
> > images)?
> In 'linked' mode, PTGui still generates control points for all images,
> but it ignores the cps between the linked images. And the cp's of all
> linked images are taken together for optimization.
[Non-text portions of this message have been removed]