4d — 4D Full-Screen Browser

← Back to the 4d application (GUI)

The 4D full-screen browser — inspect a 4D-STEM dataset

The 4D full-screen browser opens whenever you open a 4D-STEM dataset in 4d's native HDF5 format (an .h5 file, e.g. from the Library or a results panel). It fills the screen with a split view: on the left the real-space side — the scan positions and the reconstructed image — and on the right the reciprocal-space side — the individual diffraction pattern recorded at the current scan position. Simply moving the mouse across the left panel selects the nearest scan position and updates the right panel live, so you can "fly" over your specimen and watch the diffraction patterns change.

Across the top runs a control bar with four groups — scan-coordinate loading and DF/BF integration, Shadow-image calculation, iDPC calculation, and Circle positioning. The small square ? help button in the top-right corner, just below the Circle positioning panel, opens this manual page.

The window, region by region

The screenshot below shows the full-screen browser with a cryo-EM 4D-STEM dataset loaded. Each region is marked with a number; the legend that follows explains what each one is and does.

Screenshot images/screenshot-4d-fullscreenbrowser.png not found — place it in the manual’s images/ directory.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

The Image Statistics window (key A)

Pressing A opens the Image Statistics window with the average and standard-deviation maps over all diffraction patterns of the dataset — useful for checking the detector illumination, finding the bright-field disk radius, and judging overall data quality.

Screenshot images/screenshot-4d-fullscreenbrowser-average.png not found — place it in the manual’s images/ directory.

1 2 3 4 5 6 7

Keyboard shortcuts

The right mouse click (and other mouse actions)

A right mouse click anywhere in the browser opens the 4D Viewer options menu (also reachable with Ctrl + left click). It collects all the actions above in one place, with their shortcuts shown:

Other mouse actions:

• Left click (over the images) — open the floating zoom window magnifier at the clicked point; keep the button pressed to drag it around.
• Mouse wheel over the left panel — zoom the scan-position plot (1–20×), centred on the cursor; the reconstruction under the dots zooms along.
• Middle click / Alt + left drag — pan the view when zoomed in.
• Hover over the diffraction pattern — read out the pixel value under the cursor.

The reconstructions: DF, BF, Shadow and iDPC

All four reconstructions turn the 4D dataset (a 2D diffraction pattern for every 2D scan position) into a 2D real-space image displayed in the left panel. They differ in which part of each pattern they use and how they turn it into contrast. The BF disk and DF annulus are defined once, in the Circle positioning panel, and shared by all of them.

Circle positioning — defining the detector regions

Toggle the overlays with Show/Hide Circles (key C) and set, in detector pixels: the BF Radius (blue disk — the unscattered "bright-field" beam), and the DF Inner/DF Outer radii (dark-yellow annulus — the scattered "dark-field" signal). Center X/Y shifts both overlays together to compensate a de-centred diffraction pattern; with the overlay visible, the arrow keys nudge the center in 0.1-pixel steps. Use the average map of the Image Statistics window to measure the disk radius and center precisely.

BF and DF images — virtual detectors

Calculate BF and Calculate DF emulate the classical STEM detectors: for every scan position, the diffraction pattern is integrated over the BF disk (BF image) or over the DF annulus (DF image), and the integral becomes the grey value of that scan position in the reconstructed image. Bright-field contrast is dominated by absorption/phase effects of the transmitted beam; dark-field contrast highlights strongly scattering (e.g. heavier or thicker) regions. The calculation runs progressively — the image sharpens as more positions are evaluated.

Shadow images — magnified projection from overlapping disks

The shadow (or "shadowgraph") reconstruction exploits the fact that with a defocused or convergent probe, each bright-field disk is a small shadow image of the specimen area around the probe position. The algorithm places every pattern's selected region — mean-subtracted — at its scan position on a large canvas and sums the overlaps:

• CF Factor — the conversion factor from scan coordinates to canvas pixels. It sets the magnification of the shadow image: a larger CF spreads the scan positions further apart, producing a larger, more magnified image. The slider stops at the value where the resulting image would exceed 6000 × 6000 pixels.
• Rotation — rotates each pattern about its center to compensate the rotation between scan axes and detector axes. When the rotation (and CF) are correct, features from overlapping disks reinforce instead of blurring.
• Live preview — while you drag either slider, a small white-framed window in the image center shows the reconstruction live. Its size adapts so one preview frame is computed per second; when you pause the slider, increasingly larger frames are computed until the whole image is covered, and when you release the slider the full image is calculated.
• Calculate BF / DF / FullField Shadow Image — run the full reconstruction using the BF disk, the DF annulus, or the entire pattern, respectively.

iDPC — integrated differential phase contrast

Differential phase contrast measures the deflection of the beam by the local electric field of the specimen. For every scan position, the annular region of the pattern is split into two half-circles and their mean intensities subtracted:

• DPC in X — left-half minus right-half → horizontal deflection map.
• DPC in Y — top-half minus bottom-half → vertical deflection map.
• DPC-Rotation — rotates the half-circle split line to compensate the scan-vs-detector rotation; adjust it until edges in the DPC maps look symmetric rather than sheared.
• iDPC map — treats (DPC_x, DPC_y) as a gradient field and integrates it in Fourier space (F(k) = i(k_xF_x + k_yF_y) / k²). For a thin specimen the result is proportional to the projected electrostatic potential — a directly interpretable phase image with high dose efficiency.
• HP iDPC map — the same, followed by a Gaussian high-pass filter (1 − e^−k²/2σ², with σ = HP Width in reciprocal pixels) that removes the slowly varying background the 1/k² integration amplifies.