All, I'm not sure what's gone wrong with this example:
library(ggplot2)
par(ask = TRUE)
alphaVals <- seq(0.1, 1, 0.1)
for(alpha in alphaVals) print(qplot(x = 1:100, y = 1:100) + geom_rect(xmin = 20, xmax = 70, ymin = -Inf, ymax = Inf, alpha = alpha, fill = 'grey50'))
You can see that from alpha equals 0 to around 0.2 I get some transparency, but after that it's just gone. I've never had problems setting the alpha scale of a ggplot2 layer before.
What am I doing wrong here?
The problem here is that ggplot2 is drawing the rectangle 100 times in the same location. Thus, 100 stacked transparent shapes appear as a single opaque shape. I discovered this by inspecting the pdf output with Adobe Illustrator. I have provided a possible solution below (re-written to use ggplot syntax instead of qplot). I certainly feel that this behavior is unexpected, but I'm not sure if it deserves to be called a bug.
My proposed solution involves (1) putting the rectangle data in its own data.frame, and (2) specifying the data separately in each layer (but not in the ggplot() call).
library(ggplot2)
dat = data.frame(x=1:100, y=1:100)
rect_dat = data.frame(xmin=20, xmax=70, ymin=0, ymax=100)
# Work-around solution.
p = ggplot() +
geom_point(data=dat, aes(x=x, y=y)) +
geom_rect(data=rect_dat,
aes(xmin=xmin, xmax=xmax, ymin=ymin, ymax=ymax),
alpha = 0.3, fill = "black")
ggsave("test.png", plot=p, height=5, width=5, dpi=150)
# Original version with 100 overlapping rectangles.
p2 = ggplot(dat, aes(x=x, y=y)) +
geom_point() +
geom_rect(xmin=20, xmax=70, ymin=0, ymax=100, alpha=0.01, fill="black")
ggsave("test.pdf", height=7, width=7)
An alternative is using annotate instead of geom_rect:
ggplot(dat, aes(x = x, y = y)) +
geom_point()+
annotate("rect", xmin = 20, xmax = 70, ymin = 0,
ymax = 100, fill = "black", alpha = 0.3)
Output:
Data:
dat <- structure(list(x = 1:100, y = 1:100),
class = "data.frame",
row.names = c(NA, -100L))
