Plot gradient circles

Question

I am attempting to reproduce a Stephen Few graphic with gradient circles that demonstrates the hard wired assumption that light appears from above. Here are the circles:

How can I recreate this? Drawing the circles isn't too bad but adding gradient is where I get thrown. I am thinking grid may create something more crisp but this may be a misconception I have.

Here is the start with drawing circles:

## John Fox circle function
source("http://dl.dropboxusercontent.com/u/61803503/wordpress/circle_fun.txt")

par(mar=rep(1, 4), bg = "grey80")
plot.new()

for (i in seq(0, 1, by = .2)) {
    for (j in seq(.6, 1, by = .1)) {
        circle(i, j, .5, "cm", , 1)
    }
}

Related question: How to use R to build bubble charts with gradient fills

EDIT:

Thought I'd share the results:

And here's the code.

Answer 1

With some repeated use of clip, you can get there.

# set up a blank plot
par(mar=rep(0, 4))
par(bg="#cccccc")
plot(NA,xlim=0:1,ylim=0:1)

# define a function
grad.circ <- function(centrex,centrey,radius,col,resolution) {
  colfunc <- colorRampPalette(col)
  shades <- colfunc(resolution)

  for (i in seq_along(shades) ) {
   clip(
      centrex - radius,
      centrex + radius,
      (centrey + radius) - ((i-1) * (radius*2)/length(shades)),
      (centrey + radius) - (i     * (radius*2)/length(shades))
       )
   symbols(
     centrex,
     centrey,
     circles=radius,
     bg=shades[i],
     fg=NA,
     add=TRUE,
     inches=FALSE
          )
  }
}

# call the function
grad.circ(0.5,0.5,0.5,c("black", "white"),300)

Result:

EDIT (by Tyler Rinker):

I wanted to add the rest of the code I used to replicate the image:

FUN <- function(plot = TRUE, cols = c("black", "white")) {
    plot(NA, xlim=0:1, ylim=0:1, axes=FALSE)
    if (plot) {
        grad.circ(0.5, 0.5, 0.5, cols, 300)
    }
}

FUN2 <- function(){
    lapply(1:3, function(i) FUN(,c("white", "black")))
    FUN(F)
    lapply(1:3, function(i) FUN())
}


X11(10, 4.5)
par(mfrow=c(3, 7))
par(mar=rep(0, 4))
par(bg="gray70")
invisible(lapply(1:3, function(i) FUN2()))

Answer 2

Here is a version using rasters and rasterImage:

image <- as.raster( matrix( seq(0,1,length.out=1001), nrow=1001, ncol=1001) )
tmp <- ( row(image) - 501 ) ^2 + ( col(image) - 501 )^2
image[tmp > 500^2] <- NA

image2 <- as.raster( matrix( seq(1,0, length.out=1001), nrow=1001, ncol=1001) )
image2[ tmp > 500^2 ] <- NA

image3 <- row(image) + col(image)
image3 <- image3/max(image3)
image3[tmp>500^2] <- NA
image4 <- 1-image3
image3 <- as.raster(image3)
image4 <- as.raster(image4)

plot( 0:1, 0:1, type='n', asp=1,ann=FALSE,axes=FALSE)
rect(0,0,1,1, col='grey')
rasterImage(image, 0.2, 0.2, 0.3, 0.3)
rasterImage(image2, 0.6, 0.6, 0.7, 0.7)
rasterImage(image3, 0.6, 0.3, 0.7, 0.4)
rasterImage(image4, 0.3, 0.7, 0.4, 0.8)

Other directions of shading can be made by changing the math a little.