Using a linear gradient as `fill` aesthetic

[jbengler]

Hi Teun!

I am tinkering around with this plot using an alpha-gradient as fill.

library(ggplot2)
library(grid)

df <- data.frame(
  year = 2011:2020,
  value = c(10,9,7,6,9,10,12,11,14,15)
)

my_gradient_alpha <- function(color = "red", max_alpha = 1, start_point = 0) {
  linearGradient(
    c(NA, alpha(color, max_alpha)),
    c(start_point, 1),
    x1 = unit(0, "npc"), y1 = unit(0, "npc"),
    x2 = unit(0, "npc"), y2 = unit(1, "npc")
  )
}

df |>
  ggplot(aes(year, value)) +
  geom_area(fill = my_gradient_alpha(max_alpha = 0.5)) +
  geom_line(linewidth = 1, color = "red") +
  theme_minimal()

Now, I would like to generalize this to work with the fill aesthetic.

library(ggplot2)
library(grid)

df2 <- data.frame(
  year = 2011:2020,
  value = c(10,9,7,6,9,10,12,11,14,15,
            4,5,6,4,3,4,6,7,9,10),
  id = rep(c("id1","id2"), each = 10)
)

df2 |>
  ggplot(aes(year, value, fill = id, color = id, group = id)) +
  geom_area(alpha = 0.4, position = "identity") +
  geom_line(linewidth = 1) +
  theme_minimal()

I tried this but it does not work.

df2 |>
  ggplot(aes(year, value, fill = id, color = id, group = id)) +
  geom_area(aes(
    fill = after_scale(my_gradient_alpha(color = colour, max_alpha = 0.5))
    ),
    position = "identity") +
  geom_line(linewidth = 1) +
  theme_minimal()

Is there a way to do this in current ggplot2? If not, do you see a way to implement this?

Best
Jan

[teunbrand]

Yeah the my_gradient_alpha() should be vectorised and return a list of gradients parallel to color for this to work out.

library(ggplot2)
library(grid)

df2 <- data.frame(
  year = 2011:2020,
  value = c(10,9,7,6,9,10,12,11,14,15,
            4,5,6,4,3,4,6,7,9,10),
  id = rep(c("id1","id2"), each = 10)
)

my_gradient_alpha <- function(color = "red", max_alpha = 1, start_point = 0) {
  lapply(color, function(col) {
    linearGradient(
      c(NA, alpha(col, max_alpha)),
      c(start_point, 1),
      x1 = unit(0, "npc"), y1 = unit(0, "npc"),
      x2 = unit(0, "npc"), y2 = unit(1, "npc")
    )
  })
}

df2 |>
  ggplot(aes(year, value, fill = id, color = id, group = id)) +
  geom_area(aes(
    fill = after_scale(my_gradient_alpha(color = colour, max_alpha = 0.5))
  ),
  position = "identity") +
  geom_line(linewidth = 1) +
  theme_minimal()

^{Created on 2025-03-03 with reprex v2.1.1}

[jbengler]

Wow, this is really cool.
Thanks a lot!