Reading files

import nuscope as ns
ds = ns.load(path)                 # first dataset in the file
items = ns.load_all(path)          # every dataset in the file
print(ns.formats())                # everything NUScope can read

load() accepts a local path or a globus://<collection>/<path> URL (see Globus).

What is read natively

Instrument family	File	Metadata source
Thermo Fisher / FEI (Quanta, Helios, Scios, Nova, Apreo)	.tif	TIFF tag 34682 / 34680 (INI text), or a copied .txt/.hdr sidecar
Zeiss	.tif	TIFF tag 34118 (CZ_SEM)
Hitachi (S-4800, SU8030, S-3400)	.tif + .txt	[SemImageFile] sidecar
JEOL (JSM-7900F)	.tif + .txt	$CM_ / $SM_ sidecar
NUScope files	.h5, .nxs, .zarr, .tif	embedded NUScope document (verified on read)

Data bars burned into SEM images are removed by default (crop_databar=True), and the number of rows removed is recorded in acquisition.databar_rows_removed and in the history. Pass crop_databar=False to keep them.

Through RosettaSciIO

Everything else goes through RosettaSciIO: Gatan DigitalMicrograph DM3/DM4 (with Microscope Info, DigiScan, Session Info and EELS tags mapped to the NUScope schema), Thermo Fisher Velox EMD, TIA SER/EMI, JEOL, Bruker BCF, EDAX, MRC, Quantum Detectors MIB, EMPAD, Delmic, PantaRhei and more. Large files can be opened lazily with lazy=True.

What you get

A Dataset: the array, calibrated axes, the normalized metadata, the vendor's original metadata and the provenance.

ds.axes            # [Axis(name='y', size=448, scale=396.875, units='nm'), Axis(name='x', ...)]
ds.pixel_size_nm   # 396.875
ds.signal          # image | spectrum | spectrum_image | diffraction | stack
ds.metadata        # normalized: see Metadata
ds.original_metadata
ds.provenance.lineage()

The type of data follows from the axes: an energy axis in eV makes a spectrum or spectrum image, reciprocal axes (1/nm) make diffraction or 4D-STEM data.