OMERO failed to import a large jp2 file

Hi,

We have a series of really big images, or virtual slides, of monkey brain tissue.

We tried to upload one of them (MBF Bioscience’s customized *.jp2 file; Compression Ratio:10) to our OMERO server, but it could not process the file properly. It’s been 10 days, but the webclient still shows a clock icon indicating that it’s still processing, and iviewer shows “Loading Image Data…”.

We have tried ~10x simaller images of the same type and they were processed as expected.

Is there a size limit for an individual image for OMERO??? Can you think of a potential fix for this problem?

Image010

Image013

Import Date:      2019-06-07 14:46:32

Dimensions (XY):              41200 x 32960

Pixels Type:        uint16

Pixels Size (XYZ) (µm):    -

Z-sections/Timepoints: 1 x 1

Channels:           

Red, Green

Filesize: 530,840 KB

Best regards,
Kouichi

Hi @kouichi-c-nakamura. With “10x smaller”, you mean roughly 4000 x 3000? The cut-off for OMERO to generate an internal pyramid of sub-resolutions is 3000 x 3000 so this is likely what’s occurring. The clock implies that this pyramid for a given image has not yet been generated. Do you see any errors or alternatively on-going processing in the var/log/Processor-0.log file? Has there been significant CPU usage on the server for those 10 days? ~Josh

Thanks, Josh.

The largest jp2 we have in OMERO now is 9268 x 10040. “10x smaller” was wrong, but this one is 3–4 times smaller. We’ll look into other things you mentioned.

Best
Kouichi

We’ve encountered another image size issue with Bio-Formats. Can I open a large jpeg 2000 file with BioFormats and ImgLib2 from Java API?

Do you think that these two are related? In other words, do you think that OMERO is dependent on Bio-Formats in opening jp2 files and the issue on the Bio-Formats side is more fundamental? Is there a way to separate these out?

Best,
Kouichi

Bio-Formats is being used by OMERO to read the files. The other thread mentioned has 2 different issues, the image size issue is related to the Bio-Formats Importer in FIJI and should not cause the same issues here. The other issue with the scan line problem may also be problematic here, however I would need to test that scenario to be sure.

If the OMERO import is taking a long time however, this may be due to the fact that it is generating pyramid sub resolutions. For JP2 this should be handled natively by the format rather than having to generate the sub resolutions, a PR to address this is opened https://github.com/ome/bioformats/pull/3445 by is not yet included in a Bio-Formats release. Once this has been released it should improve the performance.

1 Like

Thanks for the update. We’ll keep an eye on it.