Loading yt-napari Layers

The yt_napari plugin comes with some helper functions to uniformly sample yt datasets and add them to an active napari viewer from a notebook. To get started, initialize an empty napari.Viewer and load in a yt dataset:

import napari
import yt

viewer = napari.Viewer()

ds = yt.load("IsolatedGalaxy/galaxy0030/galaxy0030")

yt : [INFO     ] 2024-08-28 09:43:36,644 Parameters: current_time              = 0.0060000200028298
yt : [INFO     ] 2024-08-28 09:43:36,644 Parameters: domain_dimensions         = [32 32 32]
yt : [INFO     ] 2024-08-28 09:43:36,645 Parameters: domain_left_edge          = [0. 0. 0.]
yt : [INFO     ] 2024-08-28 09:43:36,645 Parameters: domain_right_edge         = [1. 1. 1.]
yt : [INFO     ] 2024-08-28 09:43:36,645 Parameters: cosmological_simulation   = 0

To add data to napari, we need to create a new image layer. When working with generic array data, you can do this with the viewer.add_image function. Since napari works in pixel coordinates, yt_napari has a helper Scene container that can be used to sample a dataset and add it to an active viewer. To use it, first instantiate an empty Scene:

from yt_napari.viewer import Scene
yt_scene = Scene()

The new yt_scene object will have a method, add_region that will uniformly sample the dataset within a rectalinear region and add the new sample to the viewer:

left_edge = ds.domain_center - ds.arr([40, 40, 40], 'kpc')
right_edge = ds.domain_center + ds.arr([40, 40, 40], 'kpc')
res = (600, 600, 600)

yt_scene.add_region(viewer,
                    ds,
                    ("enzo", "Temperature"),
                    left_edge = left_edge,
                    right_edge = right_edge,
                    resolution= res)

yt_scene.add_region(viewer,
                    ds,
                    ("enzo", "Density"),
                    left_edge = left_edge,
                    right_edge = right_edge,
                    resolution = res)

Parsing Hierarchy : 100%|██████████████████████████████████████████████████████████| 173/173 [00:00<00:00, 31087.55it/s]
yt : [INFO     ] 2024-08-28 09:43:36,691 Gathering a field list (this may take a moment.)

from napari.utils import nbscreenshot
nbscreenshot(viewer)

Now if we add another layer covering a different extent and at a different resolution, it will be properly scaled and translated so that it aligns with the existing layers:

left_edge = ds.domain_center - ds.arr([20, 20, 20], 'kpc')
right_edge = ds.domain_center + ds.arr([10, 10, 10], 'kpc')
yt_scene.add_region(viewer,
                    ds,
                    ("enzo", "TotalEnergy"),
                    left_edge = left_edge,
                    right_edge = right_edge,
                    resolution=(200, 200, 200))

from napari.utils import nbscreenshot
nbscreenshot(viewer)

The add_region method also accepts any keyword argument that viewer.add_image accepts, so that you can also control, for example, the colormap, contrast limits (the colormap limits) and opacity of a layer:

viewer.layers.clear() # first clear existing layers

left_edge = ds.domain_center - ds.arr([10, 10, 10], 'kpc')
right_edge = ds.domain_center + ds.arr([10, 10, 10], 'kpc')
yt_scene.add_region(viewer,
                    ds,
                    ("enzo", "Density"),
                    left_edge = left_edge,
                    right_edge = right_edge,
                    resolution=(600, 600, 600),
                    colormap="magma",
                    contrast_limits=(3.4, 6.4))

yt_scene.add_region(viewer,
                    ds,
                    ("enzo", "Temperature"),
                    left_edge = left_edge,
                    right_edge = right_edge,
                    resolution=(300, 300, 300),
                    colormap="yellow",
                    contrast_limits=(4., 6.),
                    opacity=0.6)

viewer.dims.ndisplay = 3 # switch to 3d dispaly

from napari.utils import nbscreenshot
nbscreenshot(viewer)

In addition to 3D regions, you can also load Covering Grids.

Unlike the base YTCoveringGrid, here you specify both left and right edges and the dimensions will be calculated for the level you specify. Since we’re using a divergence field here, we specifiy num_ghost_zones=1. The contrast limits are chosen to highlight the one region of strong positive divergence:

yt_scene.add_covering_grid(viewer, ds, ('gas', 'velocity_divergence'),
                           left_edge = ds.domain_center - ds.quan(0.0075, 'code_length'),
                           right_edge = ds.domain_center + ds.quan(0.0075, 'code_length'),
                           level = 8,
                           num_ghost_zones=1, take_log=False,
                           contrast_limits = (0.000000000000033, 0.000000000000204),
                           colormap='cyan',)

cam_dict = {"center": [299.25, 299.25, 299.25],
            "zoom": 1.9292299290000003,
            "angles": [96.23211782317102, 14.5372135548309, 104.6325305851072],
            "perspective": 0.0, "mouse_pan": True, "mouse_zoom": True}

viewer.camera.update(cam_dict)
viewer.layers[0].visible = False

nbscreenshot(viewer)