[ome-users] Wrong metadata and slow performance with BigTiff

Melissa Linkert melissa at glencoesoftware.com
Mon Jun 22 22:05:56 BST 2015


Hi Jan,

Thank you for the feedback, and my apologies for the delay.

> I'd like to report two issues I noticed with reading and writing
> BigTIFF format with Bio-Formats:
> 
>  1) When using Bio-Formats Exporter in Fiji to write a z-stack to
> .btf, the dimensions get mixed up, z slices are becoming t frames

This is expected if just the .btf extension (not .ome.btf) was used.
TIFF alone has no way of distinguishing between Z and T; you will need to write
OME-TIFF to have all dimensional metadata preserved.

>  2) Reading .btf is surprisingly much slower than writing, and the
> performance on Windows is almost 10-times worse than on Mac OSX,
> similar to another issue with reading OME-Tiff files on Windows:
>   http://trac.openmicroscopy.org/ome/ticket/12930
> 
> To illustrate the issue, please use the attached macro code.
> 
> On Mac OSX, I get the following output:
> 
>   Before saving: 1000 slices, 1 frames
>   Saving took 27907 ms
>   Reading took 1132937 ms
>   After saving: 1 slices, 1000 frames
> 
> On Windows, the output is:
> 
>   Before saving: 1000 slices, 1 frames
>   Saving took 224195 ms
>   Reading took 10899405 ms
>   After saving: 1 slices, 1000 frames
> 
> I used an up-to-date Fiji with 64-bit Java 1.6.0 on both OSX and Windows.

Yes, this is the same issue as noted in http://trac.openmicroscopy.org/ome/ticket/12930.
We are testing a fix for this now:

https://github.com/openmicroscopy/bioformats/pull/1848

and I would expect the upcoming 5.1.3 release to include the fix.

Regards,
-Melissa

On Thu, Jun 18, 2015 at 09:07:31AM +0200, Jan Eglinger wrote:
> Dear OME team,
> 
> I'd like to report two issues I noticed with reading and writing
> BigTIFF format with Bio-Formats:
> 
>  1) When using Bio-Formats Exporter in Fiji to write a z-stack to
> .btf, the dimensions get mixed up, z slices are becoming t frames
> 
>  2) Reading .btf is surprisingly much slower than writing, and the
> performance on Windows is almost 10-times worse than on Mac OSX,
> similar to another issue with reading OME-Tiff files on Windows:
>   http://trac.openmicroscopy.org/ome/ticket/12930
> 
> To illustrate the issue, please use the attached macro code.
> 
> On Mac OSX, I get the following output:
> 
>   Before saving: 1000 slices, 1 frames
>   Saving took 27907 ms
>   Reading took 1132937 ms
>   After saving: 1 slices, 1000 frames
> 
> On Windows, the output is:
> 
>   Before saving: 1000 slices, 1 frames
>   Saving took 224195 ms
>   Reading took 10899405 ms
>   After saving: 1 slices, 1000 frames
> 
> I used an up-to-date Fiji with 64-bit Java 1.6.0 on both OSX and Windows.
> 
> 
> Jan

> file_name = "test.btf";
> tmp = getDirectory("temp");
> newImage("BigTIFF_test", "8-bit black", 1000, 1000, 1000);
> getDimensions(w, h, c, slices_before, frames_before);
> print("Before saving: " + slices_before + " slices, " + frames_before + " frames");
> tic = getTime();
> run("Bio-Formats Exporter", "save=" + tmp + file_name +" compression=Uncompressed");
> tac = getTime();
> print("Saving took " + (tac-tic) + " ms");
> run("Bio-Formats Importer", "open=" + tmp + file_name +" color_mode=Default view=Hyperstack stack_order=XYCZT");
> toc = getTime();
> print("Reading took " + (toc-tac) + " ms");
> 
> getDimensions(w, h, c, slices_after, frames_after);
> print("After saving: " + slices_after + " slices, " + frames_after + " frames");

> _______________________________________________
> ome-users mailing list
> ome-users at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users




More information about the ome-users mailing list