hugin-devs team mailing list archive

[KFJ <791473@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

Hi all!

I've stumbled upon a strange bug while using hsi, but I think it's not
an hsi bug. If I have an image in a pto which has lens correction
coefficients differnt from zero and try to reverse-transform image
coordinates to pano coordinates, this produces (nan, nan) as output for
the image origin. All other coordinate values seem to work just fine,
and if the lens correction coefficients are all zero, the origin
transforms correctly as well. A very small value for a, b or c will
suffice to produce the bug.

I failed to produce the bug using pano-trafo with coordinates of the
image center, but then the transform never came out precisely at the
pano center, either, so maybe I just didn't manage to find the proper
parameters. I suspect the bug may be in hugin's wrapping code where the
transformations are made into objects and the shift to image-center =
0,0 is made, and I'm half certain it wasn't there a few months ago.

my system info is

Betriebssystem: Linux 2.6.38-8-generic-tuxonice i686
Architektur: 32 bit
Freier Speicher:  -1831376 kiB [sic]

Hugin
Version: Pre-Release 2011.1.0.b1083a690f94
Ressourcen-Pfad: /usr/local/share/hugin/xrc/
Datenpfad: /usr/local/share/hugin/data/

hugin is self-built - obviously with BUILD_HSI=ON.

Find attached a compressed tar file with a pto, an image and a python
script which I used to produce the error. I got this output on my
machine:

>python sad.py
image 0: ( 0.000001 , 0.000001 ) -> ( 0.000001 , 0.000001 ) success = True
image 1: ( 0.000001 , 0.000001 ) -> ( 0.000001 , 0.000001 ) success = True
image 0: ( 0.000000 , 0.000000 ) -> ( 0.000000 , 0.000000 ) success = True
image 1: ( 0.000000 , 0.000000 ) -> ( nan , nan ) success = True

Kay

** Affects: hugin
     Importance: Undecided
         Status: New


** Tags: correction lens nan reverse transform

-- 
You received this bug notification because you are a member of Hugin
Developers, which is subscribed to Hugin.
https://bugs.launchpad.net/bugs/791473

Title:
  reverse transform of (0,0) yields nan with a,b, or c != 0

Status in Hugin - Panorama Tools GUI:
  New

Bug description:
  Hi all!

  I've stumbled upon a strange bug while using hsi, but I think it's not
  an hsi bug. If I have an image in a pto which has lens correction
  coefficients differnt from zero and try to reverse-transform image
  coordinates to pano coordinates, this produces (nan, nan) as output
  for the image origin. All other coordinate values seem to work just
  fine, and if the lens correction coefficients are all zero, the origin
  transforms correctly as well. A very small value for a, b or c will
  suffice to produce the bug.

  I failed to produce the bug using pano-trafo with coordinates of the
  image center, but then the transform never came out precisely at the
  pano center, either, so maybe I just didn't manage to find the proper
  parameters. I suspect the bug may be in hugin's wrapping code where
  the transformations are made into objects and the shift to image-
  center = 0,0 is made, and I'm half certain it wasn't there a few
  months ago.

  my system info is

  Betriebssystem: Linux 2.6.38-8-generic-tuxonice i686
  Architektur: 32 bit
  Freier Speicher:  -1831376 kiB [sic]

  Hugin
  Version: Pre-Release 2011.1.0.b1083a690f94
  Ressourcen-Pfad: /usr/local/share/hugin/xrc/
  Datenpfad: /usr/local/share/hugin/data/

  hugin is self-built - obviously with BUILD_HSI=ON.

  Find attached a compressed tar file with a pto, an image and a python
  script which I used to produce the error. I got this output on my
  machine:

  >python sad.py
  image 0: ( 0.000001 , 0.000001 ) -> ( 0.000001 , 0.000001 ) success = True
  image 1: ( 0.000001 , 0.000001 ) -> ( 0.000001 , 0.000001 ) success = True
  image 0: ( 0.000000 , 0.000000 ) -> ( 0.000000 , 0.000000 ) success = True
  image 1: ( 0.000000 , 0.000000 ) -> ( nan , nan ) success = True

  Kay