为何HoloViews保存的Datashader绘图渲染不完整且分辨率低？

问题场景 #

尝试使用HoloViews结合Bokeh后端，将经Datashader处理的Curve保存为PNG时，出现渲染异常：

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

curve: hv.Curve = hv.Curve(__some_data_for_curve__)
hv.save(datashade(curve), output_path, backend="bokeh")

但改用Panel展示时，绘图能正常渲染：

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

curve: hv.Curve = hv.Curve(__some_data_for_curve__)
pn.serve(datashade(curve))

推测直接保存时Datashader未完成渲染，以下是几种解决方法：

解决方法 #

方法1：使用Panel的save方法替代hv.save #

Panel能完整处理Datashader的渲染生命周期，将HoloViews对象转为Panel组件后保存即可：

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

hv.extension("bokeh")
curve = hv.Curve(__some_data_for_curve__)
shaded_curve = datashade(curve)

# 转为Panel面板并保存，embed=True确保资源嵌入
pn.panel(shaded_curve).save(output_path, embed=True)

方法2：先渲染为Bokeh Figure再导出 #

通过hv.render将HoloViews对象转为Bokeh Figure，再用Bokeh原生工具导出PNG，确保渲染完成：

import holoviews as hv
from holoviews.operation.datashader import datashade
from bokeh.io import export_png

hv.extension("bokeh")
curve = hv.Curve(__some_data_for_curve__)
shaded_curve = datashade(curve)

# 渲染为Bokeh Figure
bokeh_fig = hv.render(shaded_curve, backend="bokeh")
# 导出PNG
export_png(bokeh_fig, filename=output_path)

方法3：强制Datashader生成静态渲染结果 #

设置dynamic=False关闭Datashader的动态交互渲染，直接生成静态图元：

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

hv.extension("bokeh")
curve = hv.Curve(__some_data_for_curve__)
# 生成静态渲染结果
shaded_curve = datashade(curve, dynamic=False)
hv.save(shaded_curve, output_path, backend="bokeh")

内容的提问来源于stack exchange，提问作者user20292042