[ome-devel] Tiled TIFF import bug

[Chris Lesiak]

I've recently found that Bio-Formats 4.4.7 and newer incorrectly read 
the TIFF images created by LI-COR Image Studio.  Although the images are 
read, the pixels are scrambled.  The last version to work was 4.4.6.

The problem is not limited to the Image Studio files.  Other tiled TIFF 
images can also trigger the bug.

A git bisect between v4.4.6 and v4.4.7 quickly found the problem commit:


64f33a89e3277a5c9a6d1bb6cbcd546e6f3a9b95 is the first bad commit
commit 64f33a89e3277a5c9a6d1bb6cbcd546e6f3a9b95
Author: Melissa Linkert <melissa at glencoesoftware.com>
Date:   Mon Mar 4 12:52:01 2013 -0500

     TIFF: adjust the optimal tile height for large single-tile images

     See
http://lists.openmicroscopy.org.uk/pipermail/ome-users/2013-March/003597.html
     and
http://lists.openmicroscopy.org.uk/pipermail/ome-users/2013-March/003601.html

:040000 040000 13d5049c3cb421dbc95eb6f3f76e6d4b549d73fa 
44df26899b2b7612e82bcaef719a1c89f2bfc956 M    components



I've uploaded the following test images:

cramps-tile.tif    This image (from pics-3.8.0.tar.gz) doesn't trigger 
the bug because the photometric interpretation is min-is-white.

cramps-tile-min-is-black.tif    This image is identical to 
cramps-tile.tif except I've changed the photometric interpretation to 
min-is-black.  It triggers the bug.

tiled-pyramid-half-hdr.tiff    This image generated by Image Studio also 
triggers the bug.  The image has a background of 0 with an an array of 
40 pixel diameter spots exercising the full range of a 16-bit float, 
including NaN, +Inf, and -Inf.


-- 
Chris Lesiak
Principal Design Engineer, Software
LI-COR, Inc.
chris.lesiak at licor.com

Any opinions expressed are those of the author and
do not necessarily represent those of his employer.