NOTICE! This is a static HTML version of a legacy Fiji BugZilla bug.

The Fiji project now uses GitHub Issues for issue tracking.

Please file all new issues there.

Bug 1234 - Issue with the Stitching plugin (Grid/collection stitching)
Issue with the Stitching plugin (Grid/collection stitching)
Status: NEW
Product: Fiji
Classification: Unclassified
Component: Plugins
unspecified
PC Windows
: P5 major
Assigned To: ImageJ Bugs Mailing List
: 1235
Depends on:
Blocks:
 
Reported: 2016-03-01 22:55 CST by suzel
Modified: 2016-03-02 07:10 CST
1 user (show)

See Also:

Description suzel 2016-03-01 22:55:01 CST
Hi,

When trying to use the row-by-row / right & up stitching module for a 4x2 grid, although the log file shows that the right images are read, the stitched montage only shows the first image stitched to  itself (i.e. a total of 8 copies of the first image).

Any chance there could be a problem with the stitching module?

Thank you.
Best,
Seb
Comment 1 Jan Eglinger 2016-03-02 02:20:25 CST
Do you try to stitch tiles from a series of files that Bio-Formats would recognize as a multi-series dataset?
I've observed that when working with a multi-file dataset, Bio-Formats tries to be smart and always opens the whole dataset. The Stitching plugin uses Bio-Formats in the backend, and will see the first series of the dataset even if any other file name from the dataset is provided in the TileConfiguration.txt file.

If you provide more details about the file format you're trying to stitch, I can try to give more specific advice how to work around the issue.
Comment 2 Jan Eglinger 2016-03-02 02:21:10 CST
*** Bug 1235 has been marked as a duplicate of this bug. ***
Comment 3 suzel 2016-03-02 05:34:46 CST
(In reply to Jan Eglinger from comment #1)
> Do you try to stitch tiles from a series of files that Bio-Formats would
> recognize as a multi-series dataset?
> I've observed that when working with a multi-file dataset, Bio-Formats tries
> to be smart and always opens the whole dataset. The Stitching plugin uses
> Bio-Formats in the backend, and will see the first series of the dataset
> even if any other file name from the dataset is provided in the
> TileConfiguration.txt file.
> 
> If you provide more details about the file format you're trying to stitch, I
> can try to give more specific advice how to work around the issue.

Hi Jan,

Thank you for responding to my post.

The set of images I'm working with is generated by the Multi-D acquisition module from micromanager. Specifically, the set of images consists of a total of 96 2-image stacks, corresponding to 12 separate regions for which I generate a 8 image (4x2) local 10% overlap coverage.
If I stitch the entire set of images using the 'position from files' sub-routine, the stitching work fine, except that it generates a massive montage image where the 12 separate regions are displayed on the same huge image and it quickly runs out of memory. That's why I was hoping to use the 'row-by-row' module to only select the few images that I'm interested in.
Alternatively, I tried to copy the few images I was interested in into a separate folder and run the 'position from file' module, except that each image contains the metadata from the entire 12 region stack, and that seems to cause troubles when trying to stitch and assemble only a subset of images.
I hope this makes sense.

Thanks again to taking the time to help out with this issue! Much appreciated!

Seb
Comment 4 Jan Eglinger 2016-03-02 06:41:53 CST
(In reply to suzel from comment #3)
> If I stitch the entire set of images using the 'position from files'
> sub-routine, the stitching work fine,

In 'positions from file', do you use the 'Defined by TileConfiguration' or the 'Defined by image metadata' mode?

If using a TileConfiguration.txt file worked for you, you can edit that file to include only the files of interest.
If using image metadata (which more likely if understood you correctly), a workaround might be to re-save the files to tiff *without* their metadata and then use row-by-row as you tried.

What format are you using? OME-Tiff?
Comment 5 suzel 2016-03-02 06:55:13 CST
(In reply to Jan Eglinger from comment #4)
> (In reply to suzel from comment #3)
> > If I stitch the entire set of images using the 'position from files'
> > sub-routine, the stitching work fine,
> 
> In 'positions from file', do you use the 'Defined by TileConfiguration' or
> the 'Defined by image metadata' mode?
> 
> If using a TileConfiguration.txt file worked for you, you can edit that file
> to include only the files of interest.
> If using image metadata (which more likely if understood you correctly), a
> workaround might be to re-save the files to tiff *without* their metadata
> and then use row-by-row as you tried.
> 
> What format are you using? OME-Tiff?

Yes, the format is ome.tif.
Also, using a TileConfiguration as an input gave me the same result as with the metadata.
However, I just tried to resave a  few stacks as you suggested and that seemed to work. I'll have to try on the full set of images. So I guess it will get the job done, although it's going to be painful to have to resave sets of 96 stacks every time I need to stitch :-/

Thanks though, it does help a lot!

Best,
Seb
Comment 6 suzel 2016-03-02 07:10:28 CST
(In reply to Jan Eglinger from comment #4)
> (In reply to suzel from comment #3)
> > If I stitch the entire set of images using the 'position from files'
> > sub-routine, the stitching work fine,
> 
> In 'positions from file', do you use the 'Defined by TileConfiguration' or
> the 'Defined by image metadata' mode?
> 
> If using a TileConfiguration.txt file worked for you, you can edit that file
> to include only the files of interest.
> If using image metadata (which more likely if understood you correctly), a
> workaround might be to re-save the files to tiff *without* their metadata
> and then use row-by-row as you tried.
> 
> What format are you using? OME-Tiff?

Also, and interestingly, the bio-format importer doesn't seem capable of generating a hyperstack when I loaded the original stacks (which I thought could be convenient to resave the entire set of images in one go instead of resaving them one by one). However, once resaved separately, bio-format importedd has no trouble grouping the stacks into a hyperstack. There seems to be some incompatibility between the image/stack format spitted out by Micromanager and the imageJ plugins for handling those stacks (which is odd considering that micromanager runs on imageJ).