Troubleshooting

This page collects a few known issues and solutions for common problems that users might encounter when using napari. If you are facing an error that is not covered by this page, please check existing issues on the napari repository. If you can’t find what you are looking for, please open a new issue.

Known issues

Plugin errors related to pydantic

napari depends on pydantic, a data validation library. In napari 0.6.0, we dropped support for pydantic v1, changing our minimum required pydantic version to 2.2. This will bring significant speed and functionality advantages to napari. Pydantic v2 was released nearly 2 years ago, so we feel that there has been adequate time for plugins to update (there is a detailed migration guide). However, there may be plugins that depend on pydantic that have not been updated to work with pydantic 2, but can be installed with napari 0.6.0, because it is not typical to set upper bounds on dependencies. In this case, you will get Import or Runtime errors related to pydantic. If this occurs, you will need to reach out to the developers of the plugin and ask them to update the plugin.

To do this, you can open the Plugin Manager using the Plugins menu: Plugins > Install/Uninstall Plugins... and then click the name of the plugin in the Installed Plugins list. This will take you to the plugin website, where you can look for information on opening an issue or contacting the developers. Alternately, you can install and use napari 0.5.6 (or lower) and specify "pydantic<2" when installing to continue to use such plugins.

Installed plugin doesn’t show up in Plugins menu

Installed npe2 plugins (this is most plugins!) are not loaded until you restart napari. Try restarting the viewer and see whether this resolves your issue.

Plugin options missing

If you’ve installed a plugin, restarted napari, and some of the plugin’s options are missing e.g. there is no Tools menu or you’re trying to save layers and can’t find the plugin in the listed writers, you might have an auto-converted plugin using the deprecated npe1 engine.

You can check if this is the problem by installing napari<=0.6.6 and turning off the Use npe2 adaptor setting under Preferences -> Plugins -> Use npe2 adaptor and restarting napari. If this fixes your issue, you’ll need reach out to the plugin developer to update their plugin in order to use it with napari>=0.7.0.

Learn more about this in the adapted plugin guide.

Resetting preferences and settings

Changing napari versions—updating or downgrading—within an environment can cause issues with settings. These issues can manifest as napari failing to start.

Likewise, issues with napari on multiple monitors or with window sizes are also due to settings. These issues can frequently be solved by resetting the settings back to default using the terminal command:

napari --reset

torch + rosetta + napari crashes Python

In some cases, when using napari on Mac M1 using Rosetta you may experience crashing when adding layers or using a script. If you observe this, you should run napari using the native arm64 Python interpreter (see napari#7259.

“module napari has no attribute Viewer”

This may happen when doing an editable install on top of a normal one, or when you have a local file called napari.py.

To fix this, either rename the file or remove all previous napari versions in your environment before installing the editable version.

My image renders as all black/all white

This can happen when the contrast limits are not set correctly. You can reset the contrast limits by right-clicking “contrast limits”, then clicking the “Reset” button in the advanced contrast limits widget shown.

See Adjusting contrast limits for more information on contrast limits.

PermissionError when trying to launch napari on Windows

If you have a PermissionError when trying to launch napari on Windows, it could be due to how numba deals with permissions. Try setting NUMBA_CACHE_DIR to an user-accessible location.

See napari#7288.

Mixed napari installations

Since napari 0.6.2 we have added a check for the mix of editable and non-editable napari installations in the same environment. This occurs when you install napari using pip install -e path/to/napari while napari related files are still present in the typical Python package installation directory (called site-packages).

Because of how importing in python works, two installations from even slightly different versions of napari will often lead to a crash on startup or other unexpected behavior

If you meet an exception starting from RuntimeError: Mix of local and non local installation detected. you have two options:

1. Recreate your environment and install napari from scratch; this is the safest option. But you need to know which packages you need to install in your environment. You may use pip freeze to get the list of packages installed in your environment.

2. Remove the non-editable napari installation from your environment. In the exception message you will see the path to the non-editable installation. For example Path to a napari directory: /home/czaki/.pyenv/versions/3.12.9/envs/napari_3.12/lib/python3.12/site-packages/napari. You can fix the error by manually deleting the napari directory. In some rare situations this solution may not fix the environment, so you may need to recreate, as described in the first option.

Running napari on Wayland with Nvidia cards

The Nvidia driver is not yet fully ready for Wayland. We found that on computers with Nvidia cards and a Wayland-based desktop environment, napari may fail to start with errors such as this:

OpenGL.error.GLError: GLError(
    err = 1280,
    baseOperation = glGetString,
    cArguments = (GL_VERSION,),
    result = b'OpenGL ES 3.2 NVIDIA 580.126.09'
)

The problem can be worked around by forcing Qt to use X11 (via XWayland) and PyOpenGL to use GLX, by setting the following environment variables:

QT_QPA_PLATFORM=xcb
PYOPENGL_PLATFORM=glx

Important

This workaround needs XWayland (the X11 compatibility layer) installed and running, since xcb connects to an X server rather than to Wayland directly. Most desktop environments ship and start it by default; if yours doesn’t, forcing xcb fails with Could not load the Qt platform plugin "xcb". Install it via your package manager (e.g. xwayland on Debian/Ubuntu, xorg-xwayland on Arch) and start a new session.

When napari detects an Nvidia proprietary driver and a reachable X server, it sets the aforementioned environment variables automatically at startup, so launching via the napari command or a Python script usually works without any action on your part. napari only sets these variables for the current session if you haven’t, so your own values are always respected.

This automatic fix only works if napari is imported before any Qt application is created. If a Qt application already exists (most commonly in IPython or Jupyter when %gui qt runs before import napari), napari can’t apply the workaround and prints a warning. In that case, set the variables yourself before starting the IPython or Jupyter session:

QT_QPA_PLATFORM=xcb PYOPENGL_PLATFORM=glx ipython
QT_QPA_PLATFORM=xcb PYOPENGL_PLATFORM=glx jupyter lab