ProbeFlow

ProbeFlow is a desktop tool for scanning tunnelling microscopy (STM) and related SPM data. It browses folders of scans and spectra, applies routine image corrections, measures with ROIs and FFT tools, and exports figures or data with enough provenance to understand how they were produced — all while keeping the physical calibration (nm, pA, V) that plain bitmaps lose. It reads Createc, Nanonis, and RHK files, and includes a command-line interface for scripting and batch work.

ProbeFlow 1.0 release candidate. Version 1.0.0rc1 is the candidate for the first standalone desktop release. Raw microscope files are treated as read-only — processing and exports always write new files.

Get started

Standalone downloads

No Python setup is required for the desktop builds:

• macOS 15+ on Apple silicon: download ProbeFlow as a .dmg
• Windows 10/11 on x64: download the ProbeFlow .exe installer

These release-candidate builds are unsigned, so macOS or Windows may ask you to confirm that you trust the download. Checksums and installation notes are available on the ProbeFlow 1.0.0 RC 1 release page.

Install from source

Python 3.11 or newer is required.

git clone https://github.com/SPMQT-Lab/ProbeFlow.git
cd ProbeFlow
python -m pip install -e .
probeflow gui

Then, in the app:

1. Open folder… and point it at a directory of .dat, .sxm, .sm4, or spectroscopy files — every supported file appears as a thumbnail.
2. Double-click a scan to open the image viewer.
3. Fix the usual artifacts: Processing → STM scan-line background… (Ctrl+Alt+B) or plane subtraction (Ctrl+Shift+B), previewing the fit before applying.
4. Draw an ROI or line profile, take measurements from the Measure tab, or open the FFT viewer (Ctrl+Shift+F).
5. Export the image, profile, table, or processed scan — most exports carry a JSON provenance sidecar recording exactly how they were made.

Everything in the viewer is also reachable from the search box (Ctrl+K) — type a few letters of what you want ("background", "profile", "fft").

The step-by-step walkthrough is in the GUI guide; batch and scripting workflows are in the command-line guide.

What it does

• Browse scans and spectra as thumbnails — channels, colormaps, sorting, and a bias filter; raw files are never modified.
• Process — row alignment, bad-line repair, background subtraction, smoothing/high-pass, edge detection and masks, Fourier filters, TV denoising, geometry transforms. Every step is recorded so an export can be reproduced.
• Measure — rectangle/ellipse/polygon/freehand/line/point ROIs, line profiles and periodicity, ROI statistics, step heights, distances and angles, feature maxima and point statistics (pair correlation g(r), nearest-neighbour spacing, density).
• FFT tools — magnitude and radial profile with q in nm⁻¹, draggable reciprocal-lattice grids, affine drift correction, mains-pickup notching, inverse-FFT reconstruction, and n-fold symmetrization.
• Spectroscopy — single traces, overlays, and waterfalls with smoothing, derivatives, and normalization on derived display data.
• Convert and export — Createc .dat → Nanonis .sxm, PNG, or NumPy .npy bundles; export PNG/PDF/CSV/JSON/SXM (and Gwyddion .gwy with the optional extra), with provenance sidecars.

Supported files

Direction	File type	Use
Input	Createc .dat	STM/SPM image scan
Input	Createc .VERT	Point spectroscopy
Input	Nanonis .sxm	STM/SPM image scan
Input	Nanonis .dat	Point spectroscopy
Input	RHK .sm4	STM/SPM image scan
Output	.sxm, .npy	Converted or processed scan data
Output	.png, .pdf	Figure / image export
Output	.csv, .json	Numerical data, metadata, provenance
Output	.gwy	Optional Gwyddion export

Optional extras

python -m pip install -e ".[lattice]"    # SIFT lattice-vector extraction (OpenCV, scikit-learn)
python -m pip install -e ".[gwyddion]"   # Gwyddion .gwy writer
python -m pip install -e ".[dev]"        # test + lint tooling

Everything else works with the core install (numpy, scipy, Pillow, PySide6, matplotlib, shapely, scikit-image).

Using ProbeFlow from Python

from probeflow import load_scan, processing

scan = load_scan("scan.dat")
scan.planes[0] = processing.align_rows(scan.planes[0], method="median")
scan.planes[0] = processing.subtract_background(scan.planes[0], order=1)
scan.save("processed.sxm")
scan.save("processed.png", colormap="gray")

Documentation

• GUI guide — the full tour with screenshots
• Command-line guide — inspection, conversion, batch pipelines
• Createc .dat reader notes — format details
• Contributor notes — setup, tests, architecture boundaries
• Review and cleanup status — code-review record

Development

python -m pip install -e ".[dev,lattice]" -c constraints.txt
pre-commit install    # once per clone — runs ruff on every commit
pytest                # run the test suite
ruff check .          # lint (same pinned ruff as CI and the hook)

The package layout and architectural boundaries are described in CONTRIBUTING.md.

Acknowledgements

ProbeFlow is developed at SPMQT-Lab at The University of Queensland.

The original Createc-decoding work was written by Rohan Platts. ProbeFlow builds on that foundation with browsing, conversion, processing, ROI workflows, spectroscopy handling, FFT tools, and export provenance.

The experimental Total Variation decomposition feature is adapted from AiSurf: Automated Identification of Surface Images, developed by Quantum Materials Modelling. The ProbeFlow adaptation has not been rigorously validated and is included for testing purposes. For scientific use, please consult and cite the original AiSurf Total Variation work.