Resaving first angle as all angles.

Question

Resaving first angle as all angles.

spimager opened this issue 9 years ago · 13 comments

We are having some trouble with the resave function in Multiview Reconstruction.
I have a dataset acquired with our OpenSPIM setup which consists of one time point and 12 angles (each angle being 2.6GB, saved as ome.tiff). We're running on a Dell R820 rackmount server running Ubuntu(512GB RAM, 64 core [~2.4GHz]). The two issues we're encountering are:

The HDF5 file simply resaves the first angle 12 times, i.e. if I select angle 7 to open then angle 1 shows up again. The same issue is seen when resaving as tifs instead.
It's very slow, taking approximately 14 hours to resave these files.

The original files were:
spim_TL01_Angle0.ome.tiff
spim_TL01_Angle1.ome.tiff
spim_TL01_Angle2.ome.tiff
spim_TL01_Angle3.ome.tiff
spim_TL01_Angle4.ome.tiff
spim_TL01_Angle5.ome.tiff
spim_TL01_Angle6.ome.tiff
spim_TL01_Angle7.ome.tiff
spim_TL01_Angle8.ome.tiff
spim_TL01_Angle9.ome.tiff
spim_TL01_Angle10.ome.tiff
spim_TL01_Angle11.ome.tiff

The resaved tifs were saved as:
img_Angle0.tif
img_Angle1.tif
img_Angle2.tif
img_Angle3.tif
img_Angle4.tif
img_Angle5.tif
img_Angle6.tif
img_Angle7.tif
img_Angle8.tif
img_Angle9.tif
img_Angle10.tif
img_Angle11.tif

I have also provided the dataset.xml files here: https://www.dropbox.com/sh/nuggy9mqqzd9gne/AADsZqswV_02IyhcD9DV_E5ha?dl=0

Answer 1 · 2015-07-21T11:08:41.000Z

Could you also provide the log of the Fiji window please?

Answer 2 · 2015-07-21T11:19:14.000Z

Unfortunately, I've already closed the Fiji log window. I could resave the images again to get it back?

Answer 3 · 2015-07-21T14:14:45.000Z

sorry, yes, that would be great. Could you first segment some beads or so in each angle and send me the log for that?

Answer 4 · 2015-07-21T23:55:19.000Z

I have uploaded the log.txt for the resave to the aforementioned DropBox folder. Unfortunately, we haven't been using beads, as we have been imaging neurons using the 3DISCO method, which doesn't allow for bead embedding. I can image some beads in the coming days and send you the log for that acquisition?

Answer 5 · 2015-07-22T13:12:36.000Z

Hi, it seems to work just fine ... what exactly is the problem? How do you look at the images?

Answer 6 · 2015-07-24T06:46:42.000Z

When I open the dataset.h5 file with Big Data Viewer, I am only able to see one view. It's the same view for every angle. Even when I choose to hide all angles except for one, it's still showing the first angle, as though it has saved the first angle in place of all the other angles.

I have uploaded the dataset file to the dropbox folder I provided above. It is 14GB, though...

As a side note, other users seem to be having troubles with the ome.tiff/bio-formats file types with their SPIM data (http://openspim.org/pipermail/openspim/2015-July/000782.html) Perhaps it's related?

Answer 7 · 2015-07-24T12:05:49.000Z

Once you define the dataset (without HDF5 conversion), can you try to just display a few angles as 32bit and tell me if that works?

Answer 8 · 2015-07-24T12:06:57.000Z

It sounds like it is related. Can you try to use the ImageJ imgloader when you define the dataset, not LOCI?

Answer 9 · 2015-07-24T12:26:05.000Z

Defining the dataset with imgloader works.
If I try displaying the data the Multiview Construction, I get the following exception:
(Fiji Is Just) ImageJ 2.0.0-rc-34/1.50a; Java 1.6.0_24 [64-bit]; Linux 3.10.0-229.4.2.el7.x86_64; 2674MB of 383190MB (<1%)

java.lang.NoClassDefFoundError: Could not initialize class org.jdom2.input.sax.XMLReaders
at org.jdom2.input.SAXBuilder.(SAXBuilder.java:338)
at org.jdom2.input.SAXBuilder.(SAXBuilder.java:221)
at mpicbg.spim.data.generic.XmlIoAbstractSpimData.load(XmlIoAbstractSpimData.java:48)
at spim.fiji.plugin.queryXML.GenericLoadParseQueryXML.parseXML(GenericLoadParseQueryXML.java:787)
at spim.fiji.plugin.queryXML.GenericLoadParseQueryXML.tryParsing(GenericLoadParseQueryXML.java:671)
at spim.fiji.plugin.queryXML.GenericLoadParseQueryXML.queryXML(GenericLoadParseQueryXML.java:230)
at spim.fiji.plugin.queryXML.LoadParseQueryXML.queryXML(LoadParseQueryXML.java:83)
at spim.fiji.plugin.queryXML.LoadParseQueryXML.queryXML(LoadParseQueryXML.java:74)
at spim.fiji.plugin.queryXML.LoadParseQueryXML.queryXML(LoadParseQueryXML.java:31)
at spim.fiji.plugin.Display_View.run(Display_View.java:45)
at ij.IJ.runUserPlugIn(IJ.java:212)
at ij.IJ.runPlugIn(IJ.java:176)
at ij.Executer.runCommand(Executer.java:132)
at ij.Executer.run(Executer.java:65)
at java.lang.Thread.run(Thread.java:662)

Answer 10 · 2015-07-24T19:45:22.000Z

With which ImgLoader, the ImageJ Imgloader? Anyhow, I cannot confirm this error ... can you please update Fiji and try again?

Answer 11 · 2015-07-30T23:50:51.000Z

I tried defining the dataset with the ImageJ Opener however because their ome.tiff files, I get the Bio-Formats Import Options dialog. If I uncheck all boxes, I get the "Bioformats Series Options" where I can select which stacks I want to include. Regardless of whether I choose one stack or all of them, it simply hangs on "import preparations complete" and never defines the dataset.

Answer 12 · 2015-07-30T23:57:24.000Z

Sorry, I forgot to mention that if I select all series in the Bioformats Series Options, it will open the first stack/angle and then fail with:

Loading calibration for: /data/users/gabrielj/data/Cerebellum/spim_TL01_Angle0.ome.tiff
Could not open file: '/data/users/gabrielj/data/Cerebellum/spim_TL01_Angle0.ome.tiff'
Defining multi-view dataset failed.

Answer 13 · 2015-07-31T10:41:33.000Z

Hi @gabrieldavisjones,

this seems to be all wrong somehow. Can you please post screenshots here of how you define the dataset? There should never be any bioformats dialog ...

Thanks so much,
Stephan