如何在R中精准定位PDF内绘图？解决顶部空白问题

问题描述 #

目前通过「绘图转PNG→插入PDF的viewport」方式将多图布局放在PDF右上角，但viewport顶部出现莫名空白条，希望找到更优的PDF绘图定位方法。

问题根源 #

1. 生成PNG时par()设置的oma（外边距）和mar（内边距）给绘图添加了额外空白，转成PNG后空白被保留
2. image_ggplot()会给图片套上ggplot的默认边距，进一步放大空白
3. PDF设备默认边距也可能影响定位精度

优化方案 #

方案一：直接在PDF设备中用grid布局（推荐，矢量无损失） #

跳过PNG转换步骤，直接在PDF设备内用grid和base绘图结合的方式，精准控制多图位置，完全避免格式转换带来的空白问题。

librarian::shelf(grid)

# 打开PDF设备，设置尺寸，关闭默认边距
pdf("combined_output_opt1.pdf", width = 8.27, height = 11.69, useDingbats = FALSE)
# 清空全局默认边距
par(mar = c(0,0,0,0), oma = c(0,0,0,0))
plot.new()

# 创建右上角的viewport，精准定位
vp_topright <- viewport(
  x = 1, y = 1, 
  width = 0.4, height = 0.4, 
  just = c("right", "top")
)
pushViewport(vp_topright)

# 在viewport内设置绘图参数，仅保留必要内边距
par(
  mfrow = c(2,2),
  mar = c(3,1,1.5,2),
  pch = 16,
  lwd = 2,
  col = "darkturquoise",
  bg = "snow",
  cex.axis = 1.3,
  cex = 1.2
)

# 绘制四个子图
plot(1:10, main = "Plot 1", pch = 19, col = "darkblue", cex = 1.5)
lines(1:10, col = "red", lwd = 2)

hist(rnorm(100), main = "Histogram", col = "skyblue")

barplot(table(sample(letters[1:5], 100, replace = TRUE)), main = "Barplot", col = rainbow(5))

x <- rnorm(1000)
y <- rnorm(1000)
plot(x, y, main = "Scatterplot Example", col = rgb(0, 100, 0, 50, maxColorValue = 150))

popViewport()
dev.off()

方案二：优化现有PNG转PDF流程，消除空白 #

如果必须保留PNG转换步骤，需要清除PNG本身的空白，并去掉image_ggplot()带来的额外边距：

librarian::shelf(magick, grid)

x <- rnorm(1000)
y <- rnorm(1000)

# 生成PNG时关闭所有多余边距
png("my_new_clean.png", width = 800, height = 800, bg = "transparent")
par(
  oma = c(0,0,0,0),  # 关闭外边框
  mar = c(3,1,1.5,2), # 仅保留绘图必要内边距
  mfrow = c(2,2),
  pch = 16,
  lwd = 2,
  col = "darkturquoise",
  bg = "snow",
  cex.axis = 1.3,
  cex = 1.2
)

# 绘制四个子图
plot(1:10, main = "Plot 1", pch = 19, col = "darkblue", cex = 1.5)
lines(1:10, col = "red", lwd = 2)

hist(rnorm(100), main = "Histogram", col = "skyblue")

barplot(table(sample(letters[1:5], 100, replace = TRUE)), main = "Barplot", col = rainbow(5))

plot(x, y, main = "Scatterplot Example", col = rgb(0, 100, 0, 50, maxColorValue = 150))

dev.off()

# 读取并自动裁剪PNG的空白区域
image <- image_read("my_new_clean.png") %>%
  image_trim() 

# 按比例缩放图片
new_width <- image_info(image)$width * 0.45
new_height <- image_info(image)$height * 0.45
image_scaled <- image_scale(image, paste0(new_width, "x", new_height))

# 插入PDF，直接用grid.raster避免ggplot默认边距
pdf("combined_output_opt2.pdf", width = 8.27, height = 11.69, useDingbats = FALSE)
grid.newpage()
vp_topright <- viewport(
  x = 1, y = 1, 
  width = 0.4, height = 0.4, 
  just = c("right", "top")
)
pushViewport(vp_topright)
grid.raster(image_scaled, width = unit(1, "npc"), height = unit(1, "npc"))
popViewport()
dev.off()

方案三：用ggplot2的annotation_custom插入图片 #

如果习惯用ggplot2，可以把PDF作为空白画布，直接插入处理好的图片：

librarian::shelf(magick, ggplot2)

# 生成并裁剪干净的PNG（同方案二）
png("my_new_clean.png", width = 800, height = 800, bg = "transparent")
par(oma = c(0,0,0,0), mar = c(3,1,1.5,2), mfrow = c(2,2))
# 绘制四个子图...（代码同前）
dev.off()

# 处理图片：裁剪空白+缩放
image <- image_read("my_new_clean.png") %>% 
  image_trim() %>% 
  image_scale("360x360")
img_raster <- as.raster(image)

# 创建空白ggplot画布，将图片定位到右上角
p <- ggplot() +
  theme_void() + # 清空所有默认元素
  annotation_custom(
    rasterGrob(img_raster),
    xmin = 0.6, xmax = 1, ymin = 0.6, ymax = 1 # 用0-1坐标范围控制位置，对应画布全尺寸
  )

# 导出PDF
ggsave("combined_output_opt3.pdf", p, width = 8.27, height = 11.69, device = "pdf")

内容的提问来源于stack exchange，提问作者S1000