ggthreed - 3d geoms and stats for ggplot2

ggthreed is a collection of ggplot2 geoms which use the threed library.

What’s in the box:

• geom_threedpie() for creating 3d pie charts.
  • I am fully aware of the crimes against visualisation I am committing here.
• stat_anaglyph() for creating red-blue anaglyph images

Installation

You can install from github

# install.packages("devtools")
devtools::install_github("coolbutuseless/threed")
devtools::install_github("coolbutuseless/ggthreed")

Red/blue anaglyphs with stat_anaglyph()

Grab your glasses and lower your expectations!!

Usage

• Works with point, line and polygon geoms. Although “works” may be a strong term - the 3d effect is very weak even if you tweak all the parameters
• It might work with other geoms, but no guarantees that it produces anything worth looking at.
• parameters:
  • zoffset, zscale - for determining how z aesthetic influences position
  • red, blue - hex colours to use for red/blue
  • switch - switch position of the colours

Example red/blue anaglyphs with points.

ggplot(mtcars) +
  geom_point(aes(mpg, y = wt, z = disp), stat = 'anaglyph', alpha = 0.5, zscale = 10) +
  theme_bw()

Example of combined 3d projection and anaglyph

library(threed)

camera_to_world <- look_at_matrix(eye = c(1.5, 1.75, 3), at = c(0, 0, 0))

obj <- threed::mesh3dobj$cube %>%
  transform_by(invert_matrix(camera_to_world)) %>%
  translate_by(c(0, 0, -3)) %>%
  perspective_projection()

ggplot(obj, aes(x, y, z = z, group = element_id)) +
  geom_polygon(fill = NA, colour='black', aes(size = hidden), stat = 'anaglyph',
               zscale = 0.05, zoffset = -1.4, zinvert = FALSE) +
  scale_linetype_manual(values = c('TRUE' = "FF", 'FALSE' = 'solid')) +
  scale_size_manual(values = c('TRUE' = 0.1, 'FALSE' = 0.5)) +
  theme_void() +
  theme(legend.position = 'none') +
  coord_equal()

Animated Anaglyph

Animated Icosahedron

See vignette('animated-anaglyph', package='ggthreed')

3d pie charts with geom_threedpie()

Usage

• Requires only an x variable - which must be discrete.
• Default stat is count
• Adjustable parameters:
  • camera position
  • pie height
  • starting angle for first pie slice
  • tilt angle of pie relative to camera

Issues/Limitations

• geom_threedpie() overrides the aspect ratio of the plot it is displaying. This is a gigantic hack! It means it is not possible to use coord_fixed() etc to change the plot aspec ratio.
• The initial panel grid and x- and y-axes are still generated by the geom. Currently the only way to turn this off it to use theme_void(). Not sure how to disable this from within the geom.
• Pie slice resolution is in increments of 2 degrees.

Example - Simple Pie Chart

ggplot(mtcars) + 
  geom_threedpie(aes(x = as.factor(cyl))) + 
  theme_void() + 
  theme(legend.position = 'bottom')

Example - Facetted Pie Chart

ggplot(diamonds)  +
  geom_threedpie(aes(as.factor(cut))) +
  facet_wrap(~clarity, labeller = label_both) +
  labs(title = "Distribution of Diamond Cuts by Clarity") +
  scale_fill_brewer(name = "Cut", palette = 'Set2') +
  theme_void()

Example - Behind the Scenes

• This plot shows the polygons making up the pie.
• Triangular polygons are rendered on top, and quadrilaterals are rendered for the side.
• The threed library is used to rotate the assembled polygons and perform perspective projection.
• By plotting the polygons in order of distance from the camera (furtherest polygons first), polygons which are behind others are hidden from view.
• The pie has no bottom.
• The quadrilaterals making up the side are the same colour as the top, but darkened by 15%.

ggplot(mtcars) + 
  geom_threedpie(aes(x = as.factor(cyl)), alpha = 0.5) + 
  theme_void() + 
  theme(legend.position = 'bottom')

Example - Pie configuration

User adjustable:

• pie height
• pie tilt
• start of first pie slice