Page 2 of 2

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Mon Jul 02, 2018 4:44 pm
by Kouichi_C_Nakamura
Glad to hear about the progress, thanks a lot.

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Wed Oct 17, 2018 11:15 am
by Kouichi_C_Nakamura
Thank you for the analysis. Any progress on this issue? I tried it again, but it still does not work.

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Fri Oct 19, 2018 3:03 pm
by dgault
Im afraid at the moment we do not yet have any further fixes in place. There was an update in 5.4.6 which would allos for improved performance when cropping the image. If that is a possibility for you then it may provide a work around.

I have reopened a previously existing Trello card for this issue to keep track of it: https://trello.com/c/2GKksxIS/539-omero ... big-images

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Sat Oct 20, 2018 10:40 am
by Kouichi_C_Nakamura
Thanks for the Trello card.

This issue means that we cannot edit ROIs stored in OMERO data with local ImageJ. A workaround may be to download the whole data or to use OMERO API to directly retrieve ROI information from OMERO.

Either way, in order to update existing ROIs while keeping the image data, I may have to rely on OMERO API.

I haven't tried yet, but from my experience so far, it can be quite tricky to do.

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Tue Oct 23, 2018 9:17 am
by mtbc
Dear Kouichi,

It's certainly true that the ImageJ plugin needs adjustment to allow it to assemble images locally by fetching tile by tile rather than trying to grab a whole plane at once from the server. It's also true that the server's ROI service is deprecated and the query and update services do not adequately offer what is needed, especially at scale.

If you need any help or guidance on using the API to work with ROIs then please do feel free to start new threads here to ask, we would be happy to try to help. If you are working at scale, e.g., with tens of thousands of ROIs on images, there may be particular challenges there. Unfortunately it remains easy to write a script that looks like it should work to, say, use the update service to adjust ROIs, but in practice needs some non-obvious adjustment.

Cheers,
Mark

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Thu Nov 01, 2018 4:19 pm
by Kouichi_C_Nakamura
I've also noticed the omero.api.IRoi has been deprecated.

I've tried using Java Gateway in MATLAB http://www.openmicroscopy.org/community/viewtopic.php?f=6&t=8615 but it looks like it's stuck with the use of Java generics, which MATLAB doesn't seem to support.

What else routes do you suggest? Or is there any ways to manipulate ROIs in OMERO from MATLAB? Should I switch to Python, if it's easier?

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Fri Nov 02, 2018 8:57 am
by jburel
Hi Kouichi

You can still use the ROI service (we do not have a replacement yet)
With the actual MATLAB toolbox, it is possible to manipulate ROI
Few examples can be found at https://docs.openmicroscopy.org/omero/5 ... atlab.html or https://github.com/ome/training-scripts ... cal/matlab

Cheers

Jmarie

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Thu Nov 08, 2018 11:54 pm
by tboudier
Dear Omero devs,

I think I ran into the same issue by trying to retrieve a large image (15k x 18k) using my java code.
There was an exception
Code: Select all
requested 266147724 bytes, maximum allowed is 256000000 bytes (see Ice.MessageSizeMax)"
. My doce is simply calling getTile function :
Code: Select all
plane = rdf.getTile(securityContext, pixels, z, t, c, XMin, YMin, (int) sxfull, (int) syfull);


Is there any simple workaround or do I need to split the reading of the image in two parts ? Thanks.

Best,

Thomas

Re: Fail to open a large image in OMERO with ImageJ

PostPosted: Fri Nov 09, 2018 9:44 am
by mtbc
Dear Thomas,

Perhaps you're simply asking for too large a tile? You may well have to request multiple smaller ones and join them into the same array client-side.

Cheers,
Mark