PointerXY/Text with dynamic image source

HoloViews
1

Hi,
Let’s say I have a 3D array that I want to display different slices of (that part is fairly standard):

import numpy as np 
import xarray as xr
import holoviews as hv
hv.extension('bokeh')

ls = np.linspace(-5, 5, 100)
xx, yy,zz = np.meshgrid(ls, ls, ls)
img_np = np.sin(xx) * np.cos(yy+zz)
img_xr = xr.DataArray(data=img_np, dims=['a','b','c'], coords=[ls, ls, ls])

def plot_c_slice(c):
    return hv.Image(img_xr.sel(c=c))
dmap = hv.DynamicMap(plot_c_slice, kdims=['c']).redim.values(c=ls)
dmap

image

I’m interested in adding a PointerXY / crosshair (both are practically the same example, BTW).

That example (in the links above) is for the simple case of a ‘static’ image (2D array), yet as I’m viewing different slices I need to get the dynamic parameter (‘c’ in the example above) into cross_hair_info function. E.g., let’s say I’m interested in text displaying mean(a,b,c), where a,b are the 2D spatial coordinates of the image, and c is the slice.

How would I achieve that?

I have to admit that although I’ve looked at some of the references related to this topic, I’m feeling a bit lost…
httpXXX://holoviews.org/user_guide/Building_Composite_Objects.html
httpXXX://holoviews.org/user_guide/Live_Data.html
httpXXX://holoviews.org/user_guide/Responding_to_Events.html
httpXXX://github.com/holoviz/holoviews/issues/1184

Thanks.

2

Hi, a solution is to create a param.Parameterized class in wich you put the stream , the cross_hair_info and you create an attribute c and a function that will return your plot.

Here is the code :

from holoviews import streams
import holoviews as hv
import panel as pn
import param
import numpy as np
import xarray as xr
hv.extension('bokeh')

class Example(param.Parameterized):
    ls = np.linspace(-5, 5, 100)
    c = param.Selector(default=ls[0],objects=ls)
    xx, yy,zz = np.meshgrid(ls, ls, ls)
    img_np = np.sin(xx) * np.cos(yy+zz)
    img_xr = xr.DataArray(data=img_np, dims=['a','b','c'], coords=[ls, ls, ls])
    img = hv.Image(img_xr.sel(c=ls[0]))
    pointer = streams.PointerXY(x=0, y=0, source=img)

    def cross_hair_info(self,x, y):
        text = hv.Text(x+0.05, y, '%.3f'% np.mean([x,y,self.c]), halign='left', valign='bottom')
        return hv.HLine(y) * hv.VLine(x) * text
    
    @param.depends('c')
    def view(self):
        self.img = hv.Image(self.img_xr.sel(c=self.c))
        pointer = streams.PointerXY(x=0, y=0, source=self.img)
        return self.img*hv.DynamicMap(self.cross_hair_info, streams=[pointer])

ex = Example()

pn.Row(ex.view,pn.Param(ex.param,widgets={'c':pn.widgets.DiscreteSlider}))

I don’t know if it’s the best solution maybe someone have a better one. But It can be a great start for you.

3

Thanks AurelienSciarra.
Using this param module seems like an elegant solution, I’ll have to do some reading on that.

For future generations, those seem to be good starting places:
https://panel.holoviz.org/user_guide/Param.html
http://holoviews.org/user_guide/Dashboards.html