How to Save Bokeh Plots#

This guide shows how to save hvPlot plots created with the Bokeh backend (the default) as interactive HTML files or as static PNG images.

Tip

Not sure whether to save with Bokeh or Matplotlib? See How to Choose a Backend for Saving Plots.

Save as an interactive HTML file#

HTML is the format of choice for Bokeh plots, as it keeps all the interactive features like zooming, panning and hover tooltips:

import hvplot.pandas  # noqa

penguins = hvplot.sampledata.penguins("pandas")

plot = penguins.hvplot.scatter(
    x='bill_length_mm', y='bill_depth_mm',
    color='species', title="Penguin Bill Dimensions"
)
plot

hvplot.save(plot, 'penguins.html')

Make the HTML file work offline#

By default, the HTML file loads the BokehJS JavaScript library from a CDN, so it needs an internet connection to be displayed. The resources argument controls this:

'cdn' (default): load the JavaScript from the CDN, small file but requires internet access.
'inline': embed the JavaScript in the file, so it works offline and in airgapped environments.
A bokeh.resources.Resources object, such as bokeh.resources.CDN or bokeh.resources.INLINE.

hvplot.save(plot, 'penguins_inline.html', resources='inline')

Embedding the JavaScript makes the file much larger, around 3 MB instead of a few tens of kB:

from pathlib import Path

for filename in ['penguins.html', 'penguins_inline.html']:
    print(f"{filename}: {Path(filename).stat().st_size / 1e6:.2f} MB")

penguins.html: 0.03 MB
penguins_inline.html: 2.82 MB

Save as a PNG image#

Bokeh plots are rendered by the browser, so exporting them to PNG requires a headless browser. Since Bokeh 3.10, the simplest option is Playwright, which downloads and manages the browser for you. Install Playwright in your virtual environment (with pip or uv or conda or pixi …). Then run this command to install the Chromium browser that Playwright uses to generate static plots:

playwright install --only-shell chromium

Once set up, hvplot.save uses it automatically and removes the toolbar from the image:

hvplot.save(plot, 'penguins.png')

Note

With Bokeh older than 3.10, PNG export requires Selenium and a web driver (geckodriver or chromedriver), see the Bokeh export documentation.

Increase the PNG resolution#

hvplot.save produces one image pixel per screen pixel, so the plot above is saved as a 700x300 image. Increasing width and height gives a larger image, but the text and markers keep the same size and look smaller. To get a sharper image with the same layout, render the plot to a Bokeh object with hv.render and export it with Bokeh’s export_png and its scale_factor argument:

import holoviews as hv
from bokeh.io import export_png

export_png(hv.render(plot.opts(toolbar=None)), filename='penguins_2x.png', scale_factor=2)

'/Users/runner/work/hvplot/hvplot/doc/how_to/penguins_2x.png'

This saves a 1400x600 image of the same plot.