Skip to content

Launch VIPP

New in 0.15.0a5 — one VIPP launcher

Open VIPP on either Windows or macOS. Managed installations start in Auto; choose CPU, Prefer GPU, or Custom inside the app when needed. A CPU-only installation still needs the optional GPU dependencies and a supported environment before it can use GPU compute.

Open the installed app on macOS

The macOS package creates ~/Applications/VIPP.app. Open your user Applications folder and double-click VIPP; no terminal or separately installed Python is required. The managed CPU-only environment remains under ~/Library/vipp.

First launch may take longer while napari loads. If the package has not been installed yet, follow Install VIPP on macOS, including checksum verification and the Open Anyway steps for this unsigned, unnotarized alpha.

Use the installed shortcut on Windows

The Windows installer creates one VIPP launcher in the Start menu and, if selected during setup, on the Desktop. CPU and CUDA installations use the same name; they do not create separate compute-mode shortcuts. See installation ownership when more than one managed installation exists.

During startup

Startup shows the VIPP brand, real progress milestones, elapsed time, and retained diagnostics. The normal Auto launch has no Automatic label or compute-profile description on the splash. Compute controls and actual-run reporting remain inside VIPP.

In 0.15.0a3, the startup progress window has no native title bar. Drag its background to move it, or use the small minimize button at the top right while you work elsewhere. Other windows can cover it; minimizing or closing this progress window does not cancel startup. It closes automatically when VIPP is ready.

When automatic checking is enabled, each launch also starts a quiet update check. A previous session's check does not suppress it. Checking alone never downloads or opens an installer.

New or updated Windows installer shortcuts use the VIPP icon. Installed desktop launches also use VIPP's running-app icon; macOS retains its branded VIPP.app. Opening VIPP inside a separately launched napari session does not replace napari's branding. Existing shortcuts need an updated installer to receive these 0.15.0a3 changes.

Open the napari widget from a manual environment

Activate the environment used for installation and run:

napari

In napari, choose:

Plugins → VIPP Workflow (napari-vipp)

The VIPP dock widget has a workflow tab bar and three main work regions:

Region What it is for
Workflow tabs Keep several independent graphs, results, caches, histories, inspectors, paths, and Batch workspaces open. Switching tabs restores state without scientific recalculation.
Node library Search or browse operations, then add them to the graph.
Graph canvas Connect and arrange the visible analysis.
Inspector Edit the selected node, calculate manual nodes, and inspect output metadata, images, histograms, or tables.

The top toolbar opens examples and workflows, saves or exports the graph, starts batch processing, and exposes display/execution settings. Its CPU / Auto / Prefer GPU / Custom control is the authored compute request; new sessions default to Auto. After calculation, the toolbar summary and compact node badges show what actually ran. Auto uses exact compatible complete-pipeline timing history when both CPU and accelerated observations exist; without history it uses reviewed safe GPU defaults. Accelerated-only history makes the next global Auto run measure CPU once on the same execution surface. Auto never silently benchmarks multiple implementations. Use Prefer GPU when every reviewed eligible accelerator should run regardless of speed. Use a reviewed Custom choice or apply a Find fastest pipeline… proposal for per-node control and measurement.

VIPP's severity-aware message strip reports graph, workflow, and compute feedback. Napari's own bottom status bar reports viewer coordinates and layer information; the two surfaces have different owners and purposes.

Open a bundled example

Choose Gear menu → Open example…. The chooser opens a new workflow tab, groups complete graph templates by task, and configures their Image Source nodes to use matching bundled samples.

For a first visit choose:

Segmentation & Labels → Red-Channel Label Cleanup

This route does not require an external file or a layer opened through napari's sample menu.

Give the graph enough space

In 0.15.0a3, the workflow dock initially aims for two-thirds of the available window height. If napari's layer controls need more space, VIPP uses the maximum height those controls allow. Drag the divider between the viewer and workflow to adjust this balance; VIPP does not keep resetting it while you work.

For a long workflow, drag the VIPP Workflow dock title bar out of napari (or double-click the title bar) and maximize the floating window. Hide the node library after adding nodes and keep the inspector visible while tuning. Use graph zoom for overview; use napari layers for full-resolution image comparison.

If VIPP is missing from the Plugins menu

Check that napari is running from the same environment into which VIPP was installed:

python -c "import sys; print(sys.executable)"
python -c "import napari_vipp; print(napari_vipp.__file__)"

If either command uses a different environment, close napari, activate the correct environment, and launch it again. See report a problem if the plugin is installed but still not discovered.

Maintainer launch scripts

Repository scripts such as scripts/launch_vipp_sample.py and scripts/launch_vipp_intensity_workflow.py --list are for development, screenshots, and repeatable review. End users should normally use Gear menu → Open example… or Open inside VIPP.