Our application enables the generation of multiplexed images, where image regions show multiple "markers" in a tissue sample to demonstrate spatial patterns of marker co-occurrence.

We enable users to toggle a custom hovertemplate that shows the underlying raw array values for each marker before they are converted to RGB:

However, this becomes noticeably slower as the number of markers grows linearly; we pass the customdata template as a stacked numpy array paired with custom hovertext.

An option to compress this template array would help in increasing the speed and effectiveness of using customdata for our hovertemplate