Combination of datashade and hv.operation.apply_when results in navy background on zoom

[JSKenyon]

Software Version Info

OS: Ubuntu 24.04
Python: 3.10.16
Firefox: 133.0.3
Environment:

anyio==4.8.0
argon2-cffi==23.1.0
argon2-cffi-bindings==21.2.0
arrow==1.3.0
asttokens==3.0.0
async-lru==2.0.4
attrs==24.3.0
babel==2.16.0
beautifulsoup4==4.12.3
bleach==6.2.0
bokeh==3.6.2
certifi==2024.12.14
cffi==1.17.1
charset-normalizer==3.4.1
click==8.1.8
cloudpickle==3.1.1
colorcet==3.1.0
comm==0.2.2
contourpy==1.3.1
dask==2024.12.1
datashader==0.16.3
debugpy==1.8.12
decorator==5.1.1
defusedxml==0.7.1
exceptiongroup==1.2.2
executing==2.1.0
fastjsonschema==2.21.1
fqdn==1.5.1
fsspec==2024.12.0
h11==0.14.0
holoviews==1.20.0
httpcore==1.0.7
httpx==0.28.1
idna==3.10
importlib_metadata==8.5.0
ipykernel==6.29.5
ipython==8.31.0
isoduration==20.11.0
jedi==0.19.2
Jinja2==3.1.5
json5==0.10.0
jsonpointer==3.0.0
jsonschema==4.23.0
jsonschema-specifications==2024.10.1
jupyter-events==0.11.0
jupyter-lsp==2.2.5
jupyter_client==8.6.3
jupyter_core==5.7.2
jupyter_server==2.15.0
jupyter_server_terminals==0.5.3
jupyterlab==4.3.4
jupyterlab_pygments==0.3.0
jupyterlab_server==2.27.3
linkify-it-py==2.0.3
llvmlite==0.43.0
locket==1.0.0
Markdown==3.7
markdown-it-py==3.0.0
MarkupSafe==3.0.2
matplotlib-inline==0.1.7
mdit-py-plugins==0.4.2
mdurl==0.1.2
mistune==3.1.0
multipledispatch==1.0.0
nbclient==0.10.2
nbconvert==7.16.5
nbformat==5.10.4
nest-asyncio==1.6.0
notebook==7.3.2
notebook_shim==0.2.4
numba==0.60.0
numpy==2.0.2
overrides==7.7.0
packaging==24.2
pandas==2.2.3
pandocfilters==1.5.1
panel==1.5.5
param==2.2.0
parso==0.8.4
partd==1.4.2
pexpect==4.9.0
pillow==11.1.0
platformdirs==4.3.6
prometheus_client==0.21.1
prompt_toolkit==3.0.48
psutil==6.1.1
ptyprocess==0.7.0
pure_eval==0.2.3
pycparser==2.22
pyct==0.5.0
Pygments==2.19.1
python-dateutil==2.9.0.post0
python-json-logger==3.2.1
pytz==2024.2
pyviz_comms==3.0.4
PyYAML==6.0.2
pyzmq==26.2.0
referencing==0.36.1
requests==2.32.3
rfc3339-validator==0.1.4
rfc3986-validator==0.1.1
rpds-py==0.22.3
scipy==1.15.1
Send2Trash==1.8.3
six==1.17.0
sniffio==1.3.1
soupsieve==2.6
stack-data==0.6.3
terminado==0.18.1
tinycss2==1.4.0
tomli==2.2.1
toolz==1.0.0
tornado==6.4.2
tqdm==4.67.1
traitlets==5.14.3
types-python-dateutil==2.9.0.20241206
typing_extensions==4.12.2
tzdata==2024.2
uc-micro-py==1.0.3
uri-template==1.3.0
urllib3==2.3.0
wcwidth==0.2.13
webcolors==24.11.1
webencodings==0.5.1
websocket-client==1.8.0
xarray==2025.1.1
xyzservices==2024.9.0
zipp==3.21.0

Description of expected behavior and the observed behavior

I am attempting to dynamically shift between a datashaded plot and the raw Points when the current window contains more points than a specified threshold using hv.operation.apply_when. This works, but the background of the resulting plot becomes navy blue on zoom at the point when the plots switches from returning a datashaded version to the raw Points.

Edit: This does not happen when using rasterize instead of datashade.

Complete, minimal, self-contained example code that reproduces the issue

import holoviews as hv
import numpy as np
from holoviews.operation.datashader import datashade

hv.extension('bokeh', width=100)

points = hv.Points(np.random.multivariate_normal((0,0), [[0.1, 0.1], [0.1, 1.0]], (500000,)))

plot = hv.operation.apply_when(
    points,
    operation=lambda points: datashade(points).opts(width=800, height=800),
    predicate=lambda x: len(x) > 50000
)
plot  # Running inside a notebook.

Screenshots or screencasts of the bug in action

• I may be interested in making a pull request to address this

[JSKenyon]

The cause of this behaviour seems to stem from plotting an empty Points/Scatter object.

import holoviews as hv
from holoviews.operation.datashader import datashade

plot = datashade(hv.Scatter([])).opts(width=800, height=800)
plot

This produces:

I would have assumed that this would leave an empty canvas. Is there an easy way to change this behaviour?

[JSKenyon]

As my final input before the weekend, this seems to be related to how datashader handles the colourmap. It seems that datashader will make the entire plot area the colour of high end of the colour map when there are no points in the input. This is inconsistent with rasterize and, I would argue, not a sensible default.

[hoxbro]

I would agree with your assessment.

As a workaround, you can do something like this:

import holoviews as hv
import numpy as np
from holoviews.operation.datashader import datashade

def no_data_white_background(plot, element):
    if np.unique(element.rgb.data.values.view(np.uint32)).size == 1:
        element.rgb.data.values = np.zeros_like(element.rgb.data.values)
    return element


datashade._postprocess_hooks = [no_data_white_background]

hv.extension('bokeh', width=100)

points = hv.Points(np.random.multivariate_normal((0,0), [[0.1, 0.1], [0.1, 1.0]], (500000,)))

hv.operation.apply_when(
    points,
    operation=lambda points: datashade(points).opts(width=800, height=800),
    predicate=lambda x: len(x) > 50000
)

Screencast.From.2025-01-17.16-02-18.mp4

[JSKenyon]

Thank you @hoxbro - that works perfectly!