sine = figure[0, 0].add_line(np.column_stack([xs, np.sin(xs)]), colors="w", name="sine")
cosine = figure[0, 0].add_line(
    np.column_stack([xs, np.cos(xs)]), colors="cyan", name="cosine"
)

# vertical lines at every half period
phase = figure[0, 0].add_inf_line(
    np.arange(0, 4.5 * np.pi, np.pi / 2, dtype=np.float32),
    axis="x",
    colors="gray",
    thickness=1,
    dash_pattern="--",
    name="phase",
)

# a graphic with one color is one entry, labelled by the graphic's name
legend = Legend(
    [
        sine.create_legend_item(),
        cosine.create_legend_item(),
        phase.create_legend_item(),
    ]
)
figure.add_imgui_window(
    legend, location="floating", rect=(0.8, 0.05, 0, 0), title="legend"
)

# one marker shape per predicted cluster
kmeans = KMeans(n_clusters=3, n_init=10, random_state=0).fit(iris["data"])
markers = np.asarray(["circle", "square", "diamond"])[kmeans.labels_]

scatter = figure[0, 0].add_scatter(
    data=iris["data"][:, :2],  # sepal length and width
    sizes=12,
    cmap="tab10",
    cmap_transform=iris["target"],  # species
    markers=markers,
    alpha=0.5,
)
# qualitative colormap, span its full range so that species k always gets color k
scatter.cmap_range = (0, scatter.cmap.num_colors)

# the species entries are colored by the colormap, the cluster entries show the marker shapes
legend = Legend(
    [
        scatter.create_legend_item(
            label="iris",
            cmap_transform_labels={
                i: str(name) for i, name in enumerate(iris["target_names"])
            },
            markers_labels={
                "circle": "cluster 0",
                "square": "cluster 1",
                "diamond": "cluster 2",
            },
        )
    ]
)

sine = figure["sine"].add_line(
    np.column_stack([xs, ys]),
    cmap="matplotlib:coolwarm",
    cmap_transform=ys,
    thickness=5,
)
figure["sine"].add_imgui_window(
    ImguiColorbar(graphics=sine, title="sin(x)"), location="right", size=80
)

# a 2d gaussian blob colored by the distance of each point from the origin, with the distribution
# of those distances drawn on the colorbar
rng = np.random.default_rng(0)
points = rng.normal(0, 1, (5_000, 2)).astype(np.float32)
radius = np.linalg.norm(points, axis=1)

blob = figure["gaussian"].add_scatter(
    points,
    cmap="bids:viridis",
    cmap_transform=radius,
    sizes=4,
)

figure["gaussian"].add_imgui_window(
    ImguiColorbar(
        graphics=blob,
        title="radius",
        histogram=np.histogram(radius, bins=100),
    ),
    location="right",
    size=100,
)