PreibischLab / BigStitcher

ImgLib2/BDV implementation of Stitching for large datasets
GNU General Public License v2.0
64 stars 14 forks source link

Small feature request: parse z-planes in the manual loaders #126

Open ScientistRachel opened 1 year ago

ScientistRachel commented 1 year ago

I am trying to optimize a BigStitcher-based workflow and would really like to use the speed advantages of one of the manual loader options (vs the slow investigating file steps of the autoloader). My data is saved as a single tif file for each z-plane (and each channel, timepoint, etc.). The autoloader can parse z-planes as part of the file name. The dropdown menu currently says that functionality is experimental, but it has been working well for my data besides the slow processing inherent in investigating each z-plane separately. Would it be possible to add z-plane functionality to one or both of the manual loader options?

Thank you!

StephanPreibisch commented 1 year ago

Hi @ScientistRachel, it looks like the SimView loader already allows you to change the z-size? Did you try that?

Best, Stephan

ScientistRachel commented 1 year ago

Hi Stephan, Thank you for the suggestion! (And apologies for the delayed reply.) The issue is that, for example, we'd like to load in z-slices 0-250 from Camera 0 and slices 150-400 from Camera 1. I was initially hoping to just trick the dataset definition with some symbolic links, but that runs very slowly with the autoloader.

I was hoping not to change the original data, but for now, I will rename the raw files from Camera 1 such that 0-->400, 400-->0, etc., then define the dataset only using slices 0-250 of the renamed files, and finally flip the z-axis with a transform in BigStitcher. The tif files can always be renamed with the original numbering once a BDV hdf5 is created, if the user needs the original naming for any other purpose.