[Octave-bug-tracker] [bug #36638] imread error while loading a TIFF file

[Jordi Gutiérrez Hermoso]

Follow-up Comment #5, bug #36638 (project octave):

Upon further investigation, Matlab can't read your file either.  It only fails
silently:

http://paste.osuv.de/index.php/PFmhI/

And here's a transcript from someone trying to figure out what this tiff is:


<bofh__> uh
<bofh__> preliminaries:
<bofh__> it's using rgb48le as a pixelformat. wtf?
<bofh__> whatever the pictureformat inside it, it's uncompressed. try running
it through bzip2 -9. it shrinks to a tenth of its original filesize.
<bofh__> whatever the pictureformat inside it is*
<bofh__> libtiff can't open it because the version on the CSC machines does
not support pixelformats with 16 bits per sample, which rgb48le qualifies as.
<bofh__> ffmpeg transcodes it to a png of size ~174KB, which appears to be
blank.
<bofh__> it's broken because it's using 16bits/sample for a pixelformat, and
the headers are completely fucked and missing fields, and the fields are
severely out-of-order on top of that.


Conclusion: whatever made this TIFF made a really bad TIFF. It looks like you
made it with Aptina Imaging and it's a snapshot from a video. Looks like
devware.exe which generated this can't generate TIFF properly.

I think marking this bug as "wontfix" is now justified.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?36638>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/