Scientific Raster Visualization Plugin for QGIS
Publication-quality raster figures — styled after rasterio.show() — without leaving QGIS.
Installation · Quick Start · Features · Colormaps · Tutorial · Codebase
QGIS native symbology is designed for cartographic layer management. Generating a publication-ready scientific figure — with a perceptually uniform colormap, pointed colorbar with annotated ticks, and styled axes — requires context switching to Python (NumPy + Matplotlib + Rasterio).
RasterViz eliminates that context switch.
The plugin embeds Matplotlib's Qt5Agg backend directly inside a QGIS dialog. Every figure it produces is visually and numerically identical to what a researcher would generate with rasterio.show() — but without touching a single line of code.
# What researchers had to do before RasterViz
import rasterio, matplotlib.pyplot as plt, numpy as np
src = rasterio.open("ndvi.tif")
arr = src.read(1).astype("float32")
arr[arr == src.nodata] = np.nan
vmin, vmax = np.nanpercentile(arr, 2), np.nanpercentile(arr, 98)
fig, ax = plt.subplots(figsize=(10, 8))
im = ax.imshow(arr, cmap="RdYlGn", vmin=vmin, vmax=vmax)
cb = fig.colorbar(im, ax=ax, extend="both")
cb.set_label("NDVI")
plt.savefig("ndvi.png", dpi=300)With RasterViz: open plugin → select layer → READ DATA & RENDER → adjust → EXPORT. No scripting required.
| Mode | Description |
|---|---|
| Single Band — Continuous | Renders one raster band with a configurable colormap and stretch |
| Discrete / Classified | Maps unique gridcodes to individual colours with labelled patch legend |
| RGB Three-Band Composite | Combines three bands with independent per-channel stretch |
- Actual Min–Max — maps the true data range without clipping
- Percentile — configurable lower/upper percentile (default 2nd–98th)
- Manual Min–Max — explicit
vmin/vmaxfor reproducible cross-date figures
- 24 domain-specific custom palettes registered at startup
- Full Matplotlib library (~60 named colormaps) via the same ◀ ▶ cycler
- Inline ramp preview, Reverse toggle
- All palettes registered via
plt.colormaps— available anywhere in QGIS
- Orientation: horizontal or vertical
- End style: Both Pointed · Right Pointed · Left Pointed · Box
- Label position: top · bottom · left · right
- Independent Bold toggles for label and tick labels
- Configurable position, length, thickness, label text, label/tick sizes, tick count, tick decimal places, padding
- Orientation-aware geometry: Length = long axis, Thickness = short axis (corrected from earlier versions)
- Auto-scan gridcodes with one click
- Per-class colour (swatch + hex), editable label, decimal places
- Show / hide gridcode value toggle (
MangrovevsMangrove (1)) - Symbol shape: Box or Circle
- Alignment: left · center · right
- Label padding, multi-column layout (1–10 columns)
- Nodata colour with alpha channel support
| Format | Example |
|---|---|
| DMS | 106°49'0.0" E |
| DM | 106°49.000' E |
| D (Decimal Degree) | 106.8167° E |
| Default (UTM / Metre) | 476234.0000 |
- X-axis and Y-axis label rotation independent (0–360°)
- X and Y tick count (2–20), font size, decimal places
- Grid line style: Solid · Dashed · Dotted
- Continuous colorbar
- Discrete legend
- Map title
- Subtitle
Generates a single figure with N×M sub-maps:
- Rows and columns 1–6 each
- Per-slot: layer, band, colormap, stretch, title, colorbar fraction/pad
- Global: H/W spacing, margins, figure dimensions (inches)
- Optional overall super-title
- Assembled via
matplotlib.gridspec.GridSpec
| Format | DPI | Use case |
|---|---|---|
| PNG | 300 | Journal submission |
| TIFF | 300 | Lossless archive |
| SVG | 150 | Poster / slide (vector) |
| 150 | Print (vector) |
All palettes below are registered automatically at plugin load time via LinearSegmentedColormap.from_list(). All standard Matplotlib colormaps (viridis, plasma, RdYlGn, jet, etc.) are also available.
No additional pip install required. All packages ship with standard QGIS.
| Package | Role | Bundled |
|---|---|---|
| PyQGIS | Layer access, raster provider, GUI integration | ✅ |
| PyQt5 | Dialog, widget, layout construction | ✅ |
| NumPy | Array operations, stretch, RGBA assembly | ✅ |
| Matplotlib | Figure rendering, colormap, colorbar, tick formatting | ✅ |
os |
File path resolution | ✅ stdlib |
Matplotlib sub-modules used: pyplot, colors, ticker, patches, gridspec, lines, backends.backend_qt5agg.
- QGIS ≥ 3.0
- Python ≥ 3.6 (included with QGIS)
- NumPy and Matplotlib (bundled in QGIS)
- Plugins → Manage and Install Plugins → All
- Search RasterViz → Install Plugin
- Access via Raster → QRVIZ → RasterViz
Or install directly from the QGIS Plugin Repository.
- Download from Releases
- Plugins → Manage and Install Plugins → Install from ZIP
git clone https://github.com/Defani/RasterViz.git
# Windows (OSGeo4W)
cp -r RasterViz %APPDATA%\QGIS\QGIS3\profiles\default\python\plugins\rasterviz
# Linux
cp -r RasterViz ~/.local/share/QGIS/QGIS3/profiles/default/python/plugins/rasterviz
# macOS
cp -r RasterViz "~/Library/Application Support/QGIS/QGIS3/profiles/default/python/plugins/rasterviz"Then: Plugins → Installed → tick RasterViz.
1. Raster → QRVIZ → RasterViz
2. OPEN RASTER FILE or select loaded layer
3. Set band number
4. READ DATA & RENDER
5. Adjust colormap, stretch, colorbar — canvas updates live
6. 💾 EXPORT → choose format
Load a GeoTIFF (e.g. Sentinel-2 NDVI) into QGIS. In the plugin's Layer & Band group, select the layer or click OPEN RASTER FILE. Set Band: 1 and click READ DATA & RENDER.
The raster is read via
QgsRasterDataProvider.block(), downsampled to Max pixels (k) for GUI responsiveness, and cached as afloat64array. All subsequent changes re-render from cache — no disk I/O.
| Mode | When to use |
|---|---|
| Actual Min–Max | Pre-scaled data (NDVI −1 to 1) |
| Percentile | Most remote sensing data — clips outliers |
| Manual | Reproducible cross-date comparison |
For NDVI: Percentile, Pmin = 2, Pmax = 98.
Use ◀ ▶ to cycle. Recommended:
| Data | Palette |
|---|---|
| NDVI | NDVI_Custom or RdYlGn_Custom |
| Mangrove | Mangrove |
| Carbon / biomass | Carbon_Stock |
| SAR backscatter | SAR_Backscatter |
| Elevation | Terrain_Custom |
| General | Viridis_Custom |
In Map Geometry & Coordinates (right panel):
| Setting | Value |
|---|---|
| Format | D (Decimal Degree) |
| Coord Decimals | 4 |
| X Tick Count | 5 |
| Y Tick Count | 5 |
| X-label Rotation | 45° — prevents overlap |
| Y-label Rotation | 0° — horizontal, easier to read |
In Continuous Colorbar Layout:
| Setting | Value |
|---|---|
| Orientation | horizontal |
| End Style | Both Pointed |
| Label Text | NDVI |
| Label Position | Bottom |
| Tick Count | 5 |
| Tick Decimals | 2 |
| Bold Label | ✅ |
Scientific rationale: Crameri et al. (2020) establish that pointed colorbar extensions communicate clipped data ranges — a scientifically meaningful signal, not decoration. Intermediate tick labels allow quantitative reading without mental interpolation.
💾 EXPORT → PNG for journals (300 DPI), SVG/PDF for posters (vector).
- Select Discrete tab
- READ DATA & RENDER
- SCAN GRIDCODES
- Per class: assign colour, edit label, set decimals
- Toggle Show Gridcode as needed
- Set symbol shape, alignment, columns in Discrete Legend Layout
- Drag legend in ✋ Drag Elements mode
- Layout Series tab → set Rows and Columns
- BUILD LAYOUT GRID
- Configure each slot (layer, band, colormap, stretch, title, colorbar)
- Set spacing and figure size
- READ ALL & RENDER LAYOUT
- EXPORT LAYOUT
| Element | Function |
|---|---|
| Title field | Map title text |
| Sub field | Subtitle below title |
| Font selector | Font family for all text |
| Size spinbox | Title font size |
| 💾 EXPORT | Exports active tab — Single Map or Layout Series |
RasterViz/
├── __init__.py # 8 lines QGIS classFactory entry point
├── qrviz.py # 42 lines QRVIZPlugin — initGui, unload, run
├── dialog.py # 1,522 lines main dialog
│ ├── DiscreteClassRow per-class colour / label / decimal widget
│ ├── LayoutSlotWidget per-slot config row for layout series
│ └── QRVIZDialog main QDialog
│ ├── _build_ui() top toolbar + tabs
│ ├── _build_single_map_tab() 3-column splitter layout
│ ├── _build_layout_series_tab() left controls + right canvas
│ ├── _on_drag_press/motion/release Matplotlib mouse drag system
│ ├── _render_continuous() colormap + pointed colorbar
│ ├── _render_discrete() RGBA class map + patch legend
│ ├── _render_rgb() 3-band composite
│ ├── _draw_layout() GridSpec multi-map assembly
│ ├── _style_axes() rotation, formatters, grid
│ ├── _make_lon_formatter() DMS/DM/DD/UTM closure
│ └── export_figure() / _export_layout()
├── colormaps.py # 152 lines palette registry
│ ├── CUSTOM_PALETTES 24 LinearSegmentedColormaps
│ └── COLORMAPS master list (~60+ total)
├── metadata.txt QGIS plugin metadata
├── icon.png toolbar icon
└── LICENSE GNU GPL v3 or later
Total: 1,724 lines across 4 Python files.
Raster reading via QgsRasterDataProvider.block() — respects CRS, nodata, and QGIS-side transforms. No separate rasterio or GDAL installation needed.
Live preview caches the NumPy float64 array after first read. Parameter changes trigger fig.clf() and redraw from cache. No I/O on adjustment.
Coordinate formatters are closures returned by _make_lon_formatter() and _make_lat_formatter(), passed to ax.xaxis.set_major_formatter(mticker.FuncFormatter(...)).
Colorbar geometry: horizontal → rect = [x, y, length, thickness]; vertical → rect = [x, y, thickness, length]. Length always = long axis, Thickness = short axis.
Perceptually uniform colormaps are a scientific requirement. Crameri, Shephard & Heron (2020) demonstrate that non-uniform colormaps introduce artificial features and distort quantitative reading of data.
Pointed colorbars communicate that data extends beyond the displayed range — scientifically significant information when percentile stretch is used.
Annotated intermediate ticks allow direct quantitative reading without mental interpolation between min and max.
- Crameri, F., Shephard, G. E., & Heron, P. J. (2020). The misuse of colour in science communication. Nature Communications, 11, 5444. https://doi.org/10.1038/s41467-020-19160-7
- Rougier, N. P., Droettboom, M., & Bourne, P. E. (2014). Ten simple rules for better figures. PLOS Computational Biology, 10(9), e1003833. https://doi.org/10.1371/journal.pcbi.1003833
- Crameri, F. (2024). Choosing suitable color palettes for accessible and accurate science figures. Current Protocols. https://doi.org/10.1002/cpz1.1126
GNU General Public License v3.0 or later. See LICENSE.
Matplotlib and NumPy are distributed under the BSD License. PyQGIS and PyQt5 are distributed under the GNU GPL v2.
Alfitriansyah, D. A. (2026). RasterViz: Scientific Raster Visualization Plugin for QGIS
(Version 1.1.0) [Software]. https://github.com/Defani/RasterViz
Built with PyQGIS · NumPy · Matplotlib · PyQt5