I’ve been creating some large mosaics with big stitcher recently.
As I have images from different microscopes and the way they structure the file names and folder structure is not directly handled by Big Stitcher I have been using my own Python scripts (using the
tifffile package) to collect/convert the various files and create
tif-files that I can import in Big Stitcher.
I create one large tif-volume per tile (single channel, single timepoint, single illumination, single camera) and then create the data set.
This has been working ok, up until I started creating tiff tiles that exceed 4GB. I can import these file in plain Fiji using Bioformats. However, when trying to import them into Big Stitcher, only the first slice is read, irregardless of whether I use the Automated Loader, Manual Loader (Bioformats-based), Manual Loader (Tiff Imagej Opener). The fact that only a single slice is read with the Tiff ImageJ opener is not unexpected (https://github.com/imagej/imagej/issues/117) but I am somewhat surprised that the Bioformats-based importer works fine in Fiji but only reads the first slice in BigStitchers Bioformats-based loader.
What are my options?
I can create BigDataViewer files directly from Python but I don’t know how to import those into Big Stitcher as tiles. (This is also similar to the unanswered question by @SebT here BigStitcher loading tiles from hdf5)
I could alternatively split those large tiles into smaller, overlapping tiles < 4GB so I can import them, but then I would have to stitch them together (possibly introducing mis-registered tiles that were previously a single volume).
Any hints are appreciated.
Edited to add:
In Big Stitcher
Define Dataset (2/3) I have been choosing
CellImg (slower, larger files supported) and it didn’t make a difference.