optimal workflow for big images

[VPrevosto]

Hi everyone,
I'm interested to hear from Omero users who routinely deal with big images. In our lab, we generate large numbers of tiled multichannel confocal images. Each image is fairly large. We can manipulate them in Omero (both .insight and .web). However, loading them in Omero.figure fails (Image 'xxx.lsm' is too big for OMERO.figure). This might perhaps be fixed through further tinkering with JVM settings, although I doubt it. More generally, I'm wondering if we should add an additional steps in our process, like down-sampling images of interest and maybe converting them to LZW-compressed OME-TIFF. Another option (or a complementary one) could be to draw ROIs on said images, and batch-crop them to ROIs. I'm wondering what's the best practice here.
Any advice would be appreciated.
Thanks
Vincent

[wmoore]

Hi Vincent,

I'm afraid that OMERO.figure doesn't support tiled image viewing, so you can only view "Big" images by loading a whole plane at a time. This means that the size of images we can handle is limited by memory etc.
Currently we check if the size is bigger than 5k x 5k pixels. If your image is just a little bigger than this, then you may be able to tweak this line in figure.js to increase the limit
https://github.com/ome/figure/blob/bb2d ... 1.js#L3297

This limit is based purely on the sizeX and sizeY of the image, so compressing the data won't help.
Down-sampling would help, if you're not concerned about loss of resolution, but I don't know the best way to do this in your workflow.
One option is to draw regions of interest on the full-sized images in OMERO, then crop them using the "Images from ROIs" script to produce images smaller than 5k * 5k that can then be added to OMERO.figure.

However, if they are still bigger than approx 3k * 3k then they will still be tiled images and you will need to use a different script to produce them (see recent discussion at
viewtopic.php?f=4&t=7874).

Hope that helps,

Will.

[VPrevosto]

Many thanks for the detailed explanation. It seems that cropping images with the "Images from ROIs" script is a workable solution and fits well in our workflow.
Best,
Vincent