2 image disappeared after aligning

37 views
Skip to first unread message

Fotografia

unread,
Oct 23, 2016, 5:20:21 AM10/23/16
to hugin and other free panoramic software
Hello everyone,

 i'm trying to align a panorama with 2 rows and 2 columns, so 4 images.

I have added all the points manually (often with repeated details i have problem with auto align, this time i haven't tried and i just added the points manually), then i pressed align, the alignment was good, then i've choosed rectilinear projection and both in the preview and in the result i've lost 2 images and i'm unable to recover the full view.

I post a link to the project, the images where tif files with extension .tif, since they are really heavy i have converted them to jpeg with extension .jpg (i hope it could be possible to simply change .tif with .jpg inside the pto with a text editor).

Thank you in advance.

By the way, if i have for example image 0 (lower left), 1 (lower right), 2 (upper right), 3 (upper left), is it enought to add control points between 0-1,1-2,2-3 or also between overlapping parts of e.g. 0-2, 0-3 and so on?

Fotografia

unread,
Oct 23, 2016, 5:20:56 AM10/23/16
to hugin and other free panoramic software
I forgot to write i'm using 2016.2.0.48cb11a23351

Fotografia

unread,
Oct 23, 2016, 5:24:43 AM10/23/16
to hugin and other free panoramic software
Sorry but the link to the images is here: https://goo.gl/c6Vx5L

Frederic Da Vitoria

unread,
Oct 23, 2016, 6:42:33 PM10/23/16
to hugin-ptx
2016-10-23 11:24 GMT+02:00 Fotografia <fotogi...@gmail.com>:
Sorry but the link to the images is here: https://goo.gl/c6Vx5L


I managed to make it work by optimizing manually:
1 - select all images in first tab and reset everything.
2 - optimize for positions (incremental...), note the standard deviation value (4.59 on my test) and press Yes
3 - optimize for positions and view, check that the new standard deviation (3.08) is smaller than the previous (4.59) and if so press Yes
4 - optimize for positions and barrel, check that the new standard deviation (3.26) is smaller than the previous (3.08) and press No (I could have avoided this line, but this is to explain the procedure)
5 - optimize for positions, view and barrel, check that the new standard deviation (3.18) is smaller than the previous (3.08) and press No (I could have avoided this line, but this is to explain the procedure)
6 - optimize for everything without translation, check that the new standard deviation (0.47) is smaller than the previous (3.08) and press Yes

The result now seems correct.

I am using 2016.2.0.be8da0221960, but my procedure should work with your version of Hugin too.

Of course, this does not explain why the procedure you used did not work.

--
Frederic Da Vitoria
(davitof)

Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org

Fotografia

unread,
Oct 24, 2016, 4:26:33 AM10/24/16
to hugin and other free panoramic software
Il giorno lunedì 24 ottobre 2016 00:42:33 UTC+2, Frederic Da Vitoria ha scritto:
2016-10-23 11:24 GMT+02:00 Fotografia <fotogi...@gmail.com>:
Sorry but the link to the images is here: https://goo.gl/c6Vx5L


I managed to make it work by optimizing manually:
1 - select all images in first tab and reset everything.

Thank you this step restored all the images!
 
2 - optimize for positions (incremental...), note the standard deviation value (4.59 on my test) and press Yes
3 - optimize for positions and view, check that the new standard deviation (3.08) is smaller than the previous (4.59) and if so press Yes
4 - optimize for positions and barrel, check that the new standard deviation (3.26) is smaller than the previous (3.08) and press No (I could have avoided this line, but this is to explain the procedure)
5 - optimize for positions, view and barrel, check that the new standard deviation (3.18) is smaller than the previous (3.08) and press No (I could have avoided this line, but this is to explain the procedure)
6 - optimize for everything without translation, check that the new standard deviation (0.47) is smaller than the previous (3.08) and press Yes

The result now seems correct.

Thank you for the explanation!
Reply all
Reply to author
Forward
0 new messages