napari: a multi-dimensional image viewer for Python

Résumé fourni par la source

napari 0.9.0 Tue, Aug 25, 2026 We're happy to announce the release of napari 0.9.0! napari is a fast, interactive, multi-dimensional image viewer for Python. It's designed for browsing, annotating, and analyzing large multi-dimensional images. It's built on top of Qt (for the GUI), vispy (for performant GPU-based rendering), and the scientific Python stack (numpy, scipy). For more information, examples, and documentation, please visit our website, https://napari.org. napari follows EffVer (Intended Effort Versioning); this is a Macro release containing awesome new features, but may require dedication of some significant time when upgrading projects to use this version. Highlights Auto-labelling of viewer axes napari 0.5.0 laid the ground work for more accurate metadata handling by allowing users and plugins to set the axis_labels on layers. napari 0.9.0 finally capitalises on that work by automatically labelling viewer axes based on the axis labels of layers present in the viewer. (#9282) Now, opening a dataset with labelled axes will result in correctly labelled axes in the viewer: Additionally, when you create a new layer from an existing one using the buttons in the GUI, the new layer now inherits the axis labels of the layer(s) it was derived from. (#9293) There is still work to be done here. For example, if layers have labels that are inconsistent with each other, napari will simply ignore layers with fewer dimensions, or layers added later. But, for most use cases, layer and viewer metadata will now be much more informative! Xarray metadata is now inherited In the same vein, if you work with Xarray — common in climate, geoscience, and many places where data ships with labelled coordinates — napari now reads metadata straight from your DataArrays. When you add an xarray object to the viewer, napari will use its dimension names as axis labels, infer scale and translate from the coordinate values, and pick up units from CF-convention units attributes on coordinates (#9316). This closed an 8-year old, double-digit, nearly-as-old-as-napari-itself issue! (#14) Status bar coordinates as floats Continuing on the theme of improved metadata, when scale and/or unit metadata is set on the layer, the status bar coordinates now have increased precision, where before they were limited to just integers. This means you can have more accurate physical estimates of your data coordinates when exploring data. (#9287) Layer controls for multiple selected layers Until now, layer controls only appeared when a single layer was selected. Now, napari dynamically builds the layer controls from your selection, so when you select several layers at once you can see and use the controls that are shared between them. Pan-zoom is always available, while the layer-specific buttons still appear only when a single layer is selected. (#9318) There's also an experimental setting, Generate GUI layer controls dynamically instead of using premade panels, that makes napari use the new dynamic controls even for single layers. Take a guided tour of the viewer New to napari, or just want a quick refresher on where everything lives? There's now a guided tour, available from Help → Take a tour. The tour highlights the main areas of the viewer — the canvas, the layer list, layer controls, the viewer buttons, the dimension sliders, and the status bar — so you can get your bearings in seconds. If the viewer is empty, napari opens the built-in Balls (3D) sample data so the walkthrough has something to show (#9290). ![Screenshot of the napari viewer guided tour][../_static/images/guided-tour.png) Contributable plugin preferences Plugins can now ship their own preferences (#9308)! By declaring configurations in their napari.yaml manifest, plugins get their own settings — stored separately from napari's own settings, automatically added to and editable from the napari Preferences dialog, and accessible programmatically from Python: from napari.settings import get_plugin_settings settings = get_plugin_settings("my-plugin") settings.reader.lazy = True Read more in the Configurations Guide to learn more about how to add this new contribution to your own plugins. Adjust grid rendering with hidden layers Grid mode with hidden layers is much improved: empty grid spaces are never shown and stride operates on the full layer list, so layer grouping doesn't change when you show or hide layers (#9244). Fuzzy find in command palette Have you used our command palette yet? It's a great way to quickly access and even discover napari functionality. Now, thanks to #8661, it's easier to find functions when you don't know the exact name, or you mistyped something. You'll need to have rapidfuzz installed to make use of it. It's automatically installed with napari[all] or napari[optional]. And if the search feels a bit off, you can set just how fuzzy you want it to be in Preferences > Experimental > Fuzzy Search Threshold. 2D slicing of surfaces Ever since we added surfaces, they have been invisible in 2D slices. Now, thanks to all the work done on thick slicing, surface slices appear in 2D view (#8783). This enhancement is accompanied by support for async slicing, which should improve viewer responsiveness when slicing large, time varying surfaces, for example. ... And you can try this out yourself with common .obj surface files thanks to a new built-in reader plugin! (#9228) You should now be able to drag and drop .obj files into napari and see them instantly. Public API for auto contrast limits For a very long time, it's been possible to set automatic contrast limits updating on a layer only through the graphical user interface. This means an extra click for many workflows and poorer reproducibility. Thanks to #9271, you can now set the auto_contrast attribute on Image layers: image_layer = viewer.add_image(..., auto_contrast=True) # or image_layer.auto_contrast = True The life-changing magic of tidying up the Viewer model The napari Viewer model, for historical reasons, is a grab-bag of many attributes, which makes usage, discoverability, and code modularity and composability a major issue. We've taken some big steps towards cleaning up one of our most important namespaces by creating Canvas (#8633) and Scene (#9323) models, which cleans up (#9363) and clarifies (napari/docs#1083) many parts of the API. The old API is in extremely widespread use, so although it is silently deprecated, it will continue to work for the foreseeable future. In the meantime, the new API should be much friendlier to work with by carefully grouping related concepts and APIs. Some examples: viewer.axes and viewer.floating_axes become viewer.scene.overlays.axes and viewer.canvas.overlays.axes. From that API, you may also guess that the scale bar is on viewer.canvas.overlays.scale_bar. You can even get a list of the current overlays (both visible and invisible) with list(viewer.canvas.overlays) and list(viewer.scene.overlays). (But do note that overlays with a _ prefix are private and may change!) viewer.camera is now viewer.scene.camera. Grid mode is now accessed at viewer.canvas.grid (e.g. viewer.canvas.grid.enabled = True). This has also enabled a new public API: you can now query the canvas size (in pixels) without accessing private napari APIs! Check viewer.canvas.size! The TL;DR: # Canvas Model viewer.scale_bar -> viewer.canvas.overlays.scale_bar viewer.text_overlay -> viewer.canvas.overlays.text viewer.floating_axes -> viewer.canvas.overlays.axes viewer.grid -> viewer.canvas.grid NOW EXPOSED -> viewer.canvas.overlays.current_slice NEW -> viewer.canvas.overlay_tiling NEW -> viewer.canvas.background_color & viewer.canvas.background_color_override NEW -> viewer.canvas.size # Scene Model viewer.camera -> viewer.scene.camera viewer.axes -> viewer.scene.overlays.axes Removal of translation code Several years ago, we started working on implementing localization machinery into napari. Unfortunately, this work has been sitting unfinished and

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.