d3.histogram() takes a numeric variable, split it into bins and count the number of value per bin. See the histogram section for basic examples. It is possible to represent the histogram with a mirror effect for each group and using a curving function for smoothing: it makes the violin plots below.
Here a smoothing function is used to transformed the mirror histogram in a violin plot.
Another way to build a violin plot is to compute a kernel density estimate. This is what is done in the density plot and ridgeline plot sections. As you can see, the result is slightly different compared to above.
A few examples expanding the concepts described above to build custom violin charts.