HSI Studio

HSI Studio / Documentation / Getting Started

Getting Started

Choose a build

Web — go to hsistudio.online. Nothing to install, no account. This is the full viewer, analysis and annotation suite.

Desktop — download the installer for your platform. The desktop build adds the Dataset Library, batch processing, the ML Hub, and Kaggle integration, and it can read and overwrite files on disk in place rather than working from copies. It requires a sign-in the first time it runs; the licence is then cached locally and re-verified about once a month.

PlatformArtifacts
Windowshsi-studio-Setup-<version>.exe (installer), hsi-studio-Portable-<version>.exe
macOShsi-studio-<version>-<arch>.dmg or .zip
Linuxhsi-studio-<version>-<arch>.AppImage

The desktop build updates itself. When a new version is published you'll see an update button in the title bar; on platforms where in-place updates aren't available it opens the download page instead.

Supported input formats

FormatExtensionsNotes
ENVI.hdr + .dat / .raw / .bil / .bip / .bsqThe header and its binary companion must be loaded together
NumPy archive.npz, .npyAny 3-D array in the archive is treated as the cube
MATLAB.matv5 and v7.3 arrays
TIFF.tif, .tiffIncluding multi-page / GeoTIFF stacks
CSV.csvSpectral libraries and pixel tables
Images.png, .jpgLoaded as 3-band cubes, useful for testing
HSI Studio project.hzA saved session — cube, mask, classes, ROIs and view state

On the desktop build these extensions are also registered as file associations, so double-clicking a .hdr, .npz, .mat, .dat or .hz file opens it in HSI Studio directly.

The landing page. Drop files anywhere on it, or open one of the bundled sample datasets to try the tool without your own data.
The landing page. Drop files anywhere on it, or open one of the bundled sample datasets to try the tool without your own data.

Loading a datacube

Three ways in:

  1. Drag and drop onto the landing page. For ENVI, drop the .hdr and its binary companion at the same time — the header alone describes the data but doesn't contain it.
  2. File → Open Datacube (Ctrl/Cmd+O) on desktop.
  3. A sample dataset from the landing page. Indian Pines, Salinas and Hyperleaf are bundled so you can explore the tool without your own data.

Once a cube is loaded the landing page is replaced by the viewer.

The viewer, showing band 30 of the Indian Pines scene. Bands on the left, the datacube in the centre, the spectral plot below.
The viewer, showing band 30 of the Indian Pines scene. Bands on the left, the datacube in the centre, the spectral plot below.

Your first five minutes

  1. Scrub the bands. Scroll the mouse wheel over the image, or drag the band slider in the sidebar. Watch how contrast shifts between wavelengths.
  2. Click a pixel. The spectral plot opens showing that pixel's full reflectance curve across every band. Click more pixels to pin several curves and compare them.
  3. Switch to RGB. Press R, or toggle RGB Composite in the top bar. HSI Studio picks bands nearest 650/550/450 nm for true colour when the header carries wavelengths; you can remap any band to any channel.
  4. Paint a class. Press B for the brush and paint over a region, or press the magic wand and click — it selects pixels with a similar spectrum, with an adjustable tolerance.
  5. Export. The Save button in the top right offers the cropped cube, the mask, the current view as a PNG, a pixel-wise CSV, or a .hz project you can reopen later.

Where to go next