Using the image layer

In this document, you will learn how to use the napari image layer, including the types of images that can be displayed, and how to set properties like the contrast, opacity, colormaps and blending mode. You will also understand how to add and manipulate a variety of different types of images both from the GUI and from the console.

A simple example

You can create a new viewer and add an image in one go using the napari.view_image function, or if you already have an existing viewer, you can add an image to it using viewer.add_image. The api of both methods is the same. In these examples we’ll mainly use view_image.

A simple example of viewing an image is as follows:

import napari
from skimage import data

cells = data.cells3d()[30, 1]  # grab some data
viewer = napari.view_image(cells, colormap='magma')
from napari.utils import nbscreenshot

nbscreenshot(viewer, alt_text="Cells")

Arguments of view_image and add_image

view_image() and add_image() accept the same layer-creation parameters.

help(napari.view_image)
Help on function view_image in module napari.view_layers:

view_image(data=None, *, channel_axis=None, rgb=None, colormap=None, contrast_limits=None, gamma=1, interpolation2d='nearest', interpolation3d='linear', rendering='mip', depiction='volume', iso_threshold=0.5, attenuation=0.05, name=None, metadata=None, scale=None, translate=None, rotate=None, shear=None, affine=None, opacity=1, blending=None, visible=True, multiscale=None, cache=True, plane=None, experimental_clipping_planes=None, title='napari', ndisplay=2, order=(), axis_labels=(), show=True) -> napari.viewer.Viewer
    Create a viewer and add an image layer.
    
    Parameters
    ----------
    data : array or list of array
        Image data. Can be N >= 2 dimensional. If the last dimension has length
        3 or 4 can be interpreted as RGB or RGBA if rgb is `True`. If a
        list and arrays are decreasing in shape then the data is treated as
        a multiscale image. Please note multiscale rendering is only
        supported in 2D. In 3D, only the lowest resolution scale is
        displayed.
    channel_axis : int, optional
        Axis to expand image along.  If provided, each channel in the data
        will be added as an individual image layer.  In channel_axis mode,
        all other parameters MAY be provided as lists, and the Nth value
        will be applied to the Nth channel in the data.  If a single value
        is provided, it will be broadcast to all Layers.
    rgb : bool or list
        Whether the image is rgb RGB or RGBA. If not specified by user and
        the last dimension of the data has length 3 or 4 it will be set as
        `True`. If `False` the image is interpreted as a luminance image.
        If a list then must be same length as the axis that is being
        expanded as channels.
    colormap : str, napari.utils.Colormap, tuple, dict, list
        Colormaps to use for luminance images. If a string must be the name
        of a supported colormap from vispy or matplotlib. If a tuple the
        first value must be a string to assign as a name to a colormap and
        the second item must be a Colormap. If a dict the key must be a
        string to assign as a name to a colormap and the value must be a
        Colormap. If a list then must be same length as the axis that is
        being expanded as channels, and each colormap is applied to each
        new image layer.
    contrast_limits : list (2,)
        Color limits to be used for determining the colormap bounds for
        luminance images. If not passed is calculated as the min and max of
        the image. If list of lists then must be same length as the axis
        that is being expanded and then each colormap is applied to each
        image.
    gamma : list, float
        Gamma correction for determining colormap linearity. Defaults to 1.
        If a list then must be same length as the axis that is being
        expanded as channels.
    interpolation : str or list
        Interpolation mode used by vispy. Must be one of our supported
        modes. If a list then must be same length as the axis that is being
        expanded as channels.
    rendering : str or list
        Rendering mode used by vispy. Must be one of our supported
        modes. If a list then must be same length as the axis that is being
        expanded as channels.
    depiction : str
        Selects a preset volume depiction mode in vispy
    
        * volume: images are rendered as 3D volumes.
        * plane: images are rendered as 2D planes embedded in 3D.
    iso_threshold : float or list
        Threshold for isosurface. If a list then must be same length as the
        axis that is being expanded as channels.
    attenuation : float or list
        Attenuation rate for attenuated maximum intensity projection. If a
        list then must be same length as the axis that is being expanded as
        channels.
    name : str or list of str
        Name of the layer.  If a list then must be same length as the axis
        that is being expanded as channels.
    metadata : dict or list of dict
        Layer metadata. If a list then must be a list of dicts with the
        same length as the axis that is being expanded as channels.
    scale : tuple of float or list
        Scale factors for the layer. If a list then must be a list of
        tuples of float with the same length as the axis that is being
        expanded as channels.
    translate : tuple of float or list
        Translation values for the layer. If a list then must be a list of
        tuples of float with the same length as the axis that is being
        expanded as channels.
    rotate : float, 3-tuple of float, n-D array or list.
        If a float convert into a 2D rotation matrix using that value as an
        angle. If 3-tuple convert into a 3D rotation matrix, using a yaw,
        pitch, roll convention. Otherwise assume an nD rotation. Angles are
        assumed to be in degrees. They can be converted from radians with
        np.degrees if needed. If a list then must have same length as
        the axis that is being expanded as channels.
    shear : 1-D array or list.
        A vector of shear values for an upper triangular n-D shear matrix.
        If a list then must have same length as the axis that is being
        expanded as channels.
    affine : n-D array or napari.utils.transforms.Affine
        (N+1, N+1) affine transformation matrix in homogeneous coordinates.
        The first (N, N) entries correspond to a linear transform and
        the final column is a length N translation vector and a 1 or a
        napari `Affine` transform object. Applied as an extra transform on
        top of the provided scale, rotate, and shear values.
    opacity : float or list
        Opacity of the layer visual, between 0.0 and 1.0.  If a list then
        must be same length as the axis that is being expanded as channels.
    blending : str or list
        One of a list of preset blending modes that determines how RGB and
        alpha values of the layer visual get mixed. Allowed values are
        {'opaque', 'translucent', and 'additive'}. If a list then
        must be same length as the axis that is being expanded as channels.
    visible : bool or list of bool
        Whether the layer visual is currently being displayed.
        If a list then must be same length as the axis that is
        being expanded as channels.
    multiscale : bool
        Whether the data is a multiscale image or not. Multiscale data is
        represented by a list of array like image data. If not specified by
        the user and if the data is a list of arrays that decrease in shape
        then it will be taken to be multiscale. The first image in the list
        should be the largest. Please note multiscale rendering is only
        supported in 2D. In 3D, only the lowest resolution scale is
        displayed.
    cache : bool
        Whether slices of out-of-core datasets should be cached upon
        retrieval. Currently, this only applies to dask arrays.
    plane : dict or SlicingPlane
        Properties defining plane rendering in 3D. Properties are defined in
        data coordinates. Valid dictionary keys are
        {'position', 'normal', 'thickness', and 'enabled'}.
    experimental_clipping_planes : list of dicts, list of ClippingPlane, or ClippingPlaneList
        Each dict defines a clipping plane in 3D in data coordinates.
        Valid dictionary keys are {'position', 'normal', and 'enabled'}.
        Values on the negative side of the normal are discarded if the plane is enabled.
        title : string, optional
        The title of the viewer window. by default 'napari'.
    ndisplay : {2, 3}, optional
        Number of displayed dimensions. by default 2.
    order : tuple of int, optional
        Order in which dimensions are displayed where the last two or last
        three dimensions correspond to row x column or plane x row x column if
        ndisplay is 2 or 3. by default None
    axis_labels : list of str, optional
        Dimension names. by default they are labeled with sequential numbers
    show : bool, optional
        Whether to show the viewer after instantiation. by default True.
    
    
    Returns
    -------
    viewer : :class:`napari.Viewer`
        The newly-created viewer.

Image data and NumPy-like arrays

napari can take any numpy-like array as input for its image layer. A numpy-like array can just be a numpy array, a dask array, an xarray, a zarr array, or any other object that you can index into and when you call np.asarray on it you get back a numpy array.

The great thing about napari support of array-like objects is that you get to keep on using your favorite array libraries without worrying about any conversions as we’ll handle all of that for you.

napari will also wait until just before it displays data onto the screen to actually generate a numpy array from your data, and so if you’re using a library like dask or zarr that supports lazy loading and lazy evaluation, we won’t force you load or compute on data that you’re not looking at. This enables napari to seamlessly browse enormous datasets that are loaded in the right way. For example, here we are browsing over 100GB of lattice lightsheet data stored in a zarr file:

Multiscale images

For exceptionally large datasets napari supports multiscale images (sometimes called image pyramids). A multiscale image is a list of arrays, where each array is downsampling of the previous array in the list, so that you end up with images of successively smaller and smaller shapes. A standard multiscale image might have a 2x downsampling at each level, but napari can support any type of multiscale image as long as the shapes are getting smaller each time.

Multiscale images are especially useful for incredibly large 2D images when viewed in 2D or incredibly large 3D images when viewed in 3D. For example this ~100k x 200k pixel pathology image consists of 10 pyramid levels and can be easily browsed as at each moment in time we only load the level of the multiscale image and the part of the image that needs to be displayed:

This example had precomputed multiscale images stored in a zarr file, which is best for performance. If, however you don’t have a precomputed multiscale image but try and show a exceptionally large image napari will try and compute the multiscale image for you unless you tell it not too.

You can use the multiscale keyword argument to specify if your data is a multiscale image or not. If you don’t provide this value, then will try and guess whether your data is or needs to be a multiscale image.

Loading multichannel images

Each channel in a multichannel image can be displayed as an individual layer by using the channel_axis argument in viewer.add_image(). All the rest of the arguments to viewer.add_image() (e.g. name, colormap, contrast_limit) can take the form of a list of the same size as the number of channels.

For example, the multichannel image below has dimensions (60, 2, 256, 256) with axes ordered ZCYX (so the channel axis has an index of 1). It is loaded into napari in one line.

import napari
from skimage import data

cells = data.cells3d() #ZCYX image data

# load multichannel image in one line
viewer = napari.view_image(cells, channel_axis=1)

# load multichannel image in one line, with additional options
viewer = napari.view_image(
        cells,
        channel_axis=1,
        name=["membrane", "nuclei"],
        colormap=["green", "magenta"],
        contrast_limits=[[1000, 20000], [1000, 50000]],
        )

image: multichannel image

Viewing RGB vs luminance (grayscale) images

In this example we explicitly set the rgb keyword to be True because we know we are working with an rgb image:

viewer = napari.view_image(data.astronaut(), rgb=True)
from napari.utils import nbscreenshot

nbscreenshot(viewer, alt_text="napari viewer with the left sidebar layer controls and an image of astronaut Eileen Collins. In the layer controls, the colormap is fixed to RGB")

If we had left that keyword argument out napari would have successfully guessed that we were trying to show an rgb or rgba image because the final dimension was 3 or 4. If you have a luminance image where the last dimension is 3 or 4 you can set the rgb property to False so napari handles the image correctly.

rgb data must either be uint8, corresponding to values between 0 and 255, or float and between 0 and 1. If the values are float and outside the 0 to 1 range they will be clipped.

Working with colormaps

napari supports any colormap that is created with vispy.color.Colormap. We provide access to some standard colormaps that you can set using a string of their name. These include:

list(napari.utils.colormaps.AVAILABLE_COLORMAPS)
['PiYG',
 'blue',
 'bop blue',
 'bop orange',
 'bop purple',
 'cyan',
 'gist_earth',
 'gray',
 'gray_r',
 'green',
 'hsv',
 'inferno',
 'magenta',
 'magma',
 'plasma',
 'red',
 'turbo',
 'twilight',
 'twilight_shifted',
 'viridis',
 'yellow']

Passing any of these strings as follows to set the image colormap:

viewer = napari.view_image(data.moon(), colormap='red')

You can also access the current colormap through the layer.colormap property which returns a tuple of the colormap name followed by the vispy colormap object. You can list all the available colormaps using layer.colormaps.

It is also possible to create your own colormaps using vispy’s vispy.color.Colormap object; see its full documentation here. Briefly, you can pass Colormap a list of length 3 or length 4 lists, corresponding to the rgb or rgba values at different points along the colormap.

For example, to make a diverging colormap that goes from red to blue through black, and color a random array you can do the following:

import napari
import numpy as np
from vispy.color import Colormap

cmap = Colormap([[1, 0, 0], [0, 0, 0], [0, 0, 1]])
image = np.random.random((100, 100))

viewer = napari.view_image(image, colormap=('diverging', cmap))
from napari.utils import nbscreenshot

nbscreenshot(viewer, alt_text="napari viewer with colormap example using random data")

Note in this example how we passed the colormap keyword argument as a tuple containing both a name for our new custom colormap and the colormap itself. If we had only passed the colormap it would have been given a default name.

The named colormap now appears in the dropdown menu alongside a little thumbnail of the full range of the colormap.

Adjusting contrast limits

Each image layer gets mapped through its colormap according to values called contrast limits. The contrast limits are a 2-tuple where the second value is larger than the first. The smaller contrast limit corresponds to the value of the image data that will get mapped to the color defined by 0 in the colormap. All values of image data smaller than this value will also get mapped to this color. The larger contrast limit corresponds to the value of the image data that will get mapped to the color defined by 1 in the colormap. All values of image data larger than this value will also get mapped to this color.

For example, you are looking at an image that has values between 0 and 100 with a standard gray colormap, and you set the contrast limits to (20, 75). Then all the pixels with values less than 20 will get mapped to black, the color corresponding to 0 in the colormap, and all pixels with values greater than 75 will get mapped to white, the color corresponding to 1 in the colormap. All other pixel values between 20 and 75 will get linearly mapped onto the range of colors between black and white.

In napari you can set the contrast limits when creating an Image layer or on an existing layer using the contrast_limits keyword argument or property, respectively.

viewer = napari.view_image(data.moon(), name='moon')
viewer.layers['moon'].contrast_limits=(100, 175)
from napari.utils import nbscreenshot

nbscreenshot(viewer, alt_text="A viewer where the contrast limits have been adjusted")

Because the contrast limits are defined by two values the corresponding slider has two handles, one the adjusts the smaller value, one that adjusts the larger value.

As of right now adjusting the contrast limits has no effect for rgb data.

If no contrast limits are passed, then napari will compute them. If your data is small, then napari will just take the minimum and maximum values across your entire image. If your data is exceptionally large, this operation can be very time consuming and so if you have passed an image pyramid then napari will just use the top level of that pyramid, or it will use the minimum and maximum values across the top, middle, and bottom slices of your image. In general, if working with big images it is recommended you explicitly set the contrast limits if you can.

Currently if you pass contrast limits as a keyword argument to a layer then full extent of the contrast limits range slider will be set to those values.