Scan region in Bruker SPMLab .FLT

How an SPMLab .FLT file states the scanned area, and how the reader turns it into NeXus under the scan region convention.

Raw keys

Raw key	Meaning	How the reader uses it
OffsetX, OffsetY	Centre of the scan frame	scan_offset_value_x, _y
ScanRangeX, ScanRangeY	Width and height	scan_range_x, _y
ResolutionX, ResolutionY	Pixels per line, number of lines	scan_points_x, _y
ScanDirection	FORWARD or BACKWARD, the fast direction	Sign of the fast axis
Rotation	Rotation of the frame	scan_angle_x, not applied

A .FLT holds a single channel recorded in one direction, so the fast axis is signed. The format records no slow scan direction, so the slow axis stays bare. The [Data] block holds only the height values, ResolutionX × ResolutionY 32-bit floats; no axis coordinates are stored.

Open question. Because the slow direction is absent from the header, a bare Y here states ignorance rather than an upward scan. An SPMLab or Innova manual, or a pair of up and down scans of the same area, would settle it.

Evidence

No vendor document defining OffsetX was found, so the centre reading is graded empirical test on vendor data: a 5 µm scan and a 20 µm scan of the same sample carry identical offsets, and the small scan was located inside the large one by cross-correlation. The method and the data are described centrally.

Pair	Correlation peak / next	Measured	Centre predicts	Corner predicts
PMIS2-C8_ML2_p1_5 in …_p1_20, identical offsets	0.54 / 0.10	(10.00, 10.04) µm	(10.00, 10.00) µm	(2.50, 2.50) µm

The 5 µm scan sits in the middle of the 20 µm one, one pixel (39 nm) from the centre prediction and 10.6 µm from the corner prediction. Three further pairs of the same record, with different offsets, agree to 0.02, 0.05 and 0.7 µm. The match is found only with the rows read bottom row first, which also confirms the image orientation.

Gwyddion's spmlabf.c reads the offset as a corner, without a cited source; the test contradicts it.