John Snow's cholera clusters, visualized in 3D with rayshader and ggplot2

cholera_snow.R

#theme and ggplot derived from David Kretch, his code: https://github.com/davidkretch/london_cholera_map/blob/master/london_cholera_map.R

library(HistData)
library(ggplot2)
library(ggpointdensity)
library(rayshader)

deaths = Snow.deaths
streets = Snow.streets

themeval = theme(panel.border = element_blank(), 
                 panel.grid.major = element_blank(), 
                 panel.grid.minor = element_blank(), 
                 axis.line = element_blank(), 
                 axis.ticks = element_blank(),
                 axis.text.x = element_blank(), 
                 axis.text.y = element_blank(), 
                 axis.title.x = element_blank(), 
                 axis.title.y = element_blank(),
                 legend.key = element_blank(),
                 plot.margin = unit(c(0.5, 0, 0, 0), "cm"))
js = ggplot() + 
  geom_path(data=streets,aes(x=x, y=y, group = street), color="grey50") +
  geom_pointdensity(data = deaths,aes(x=x, y=y), size=1,adjust=0.1) +
  coord_fixed() + 
  scale_color_viridis_c() +
  theme_bw() +
  themeval 

js2 = ggplot() + 
  geom_path(data = streets, aes(x=x, y=y, group = street), color="white") +
  geom_pointdensity(data = deaths, aes(x=x, y=y), size=1, adjust=0.1) +
  coord_fixed() +
  scale_color_viridis_c() +
  theme_bw() +
  themeval 

ggheight = plot_gg(list(js,js2), multicore = TRUE, raytrace=TRUE,
                   height_aes = "color", shadow_intensity = 0.3,
                   width=8,height=7, soliddepth = -100, save_height_matrix = TRUE,
                   background = "#f5e9dc", shadowcolor= "#4f463c",windowsize=c(1000,1000))

#Add label
render_label(ggheight, "Water Pump", x=1110,y=1020,z=390, textsize = 2)

#Generate animation
for(i in 1:360) {
  render_camera(phi=20,theta=45+i,fov=70,zoom=0.25)
  render_depth(focus=0.77, 
               title_text = "John Snow's London cholera map, # of nearest neighbors", 
               title_size = 35,
               filename = glue::glue("snow{i}"))
}

For anyone else chasing down a way to reproduce this, a few tips:

1. Overall, easier to do on mac
2. Start with most recent development version (on github) of rayshader
3. lots of dependencies make Windows more challenging, like freetext, then reinstalling rgl
4. Easier to make a mp4 of this with `render_movie("john_snow.mp4", phi=20, fov=70, zoom=0.30, title_text = "John Snow's London cholera map, # of nearest neighbors", title_size = 35). This will not include the bokeh focus effect, however.
5. To make a gif, follow the directions above, which will produce 360 png files, from snow1.png to snow360.png. Unfortunately, these will be ordered in alphanumeric order, so that snow1 is followed by snow10, then snow100, snow101, etc. Not the order that you want. You can put these files into a folder, and extract the vector of files with (fs package)
   files <- fs::dir_ls("folder")
6. Then extract the numeric portion of each file name with (stringr or tidyverse package)
   num <- as.numeric(str_extract(files, "\\d+")
7. then combine these two vectors into a tibble with (tibble package)
   frame <- tibble(files, num) %>% arrange(num)
8. Then turn this into a gif with the gifski package -
   gifski::gifski(frame$files, gif_file = "snow.gif", delay = 0.04, loop = TRUE, progress = TRUE)