Read, write and join or combine images. All image functions are vectorized, meaning they operate either on a single frame or a series of frames (e.g. a collage, video, or animation). Besides paths and URLs, image_read() supports commonly used bitmap and raster object types.

  density = NULL,
  depth = NULL,
  strip = FALSE,
  coalesce = TRUE,
  defines = NULL

image_read_svg(path, width = NULL, height = NULL)

image_read_pdf(path, pages = NULL, density = 300, password = "")

image_read_video(path, fps = 1, format = "png")

  path = NULL,
  format = NULL,
  quality = NULL,
  depth = NULL,
  density = NULL,
  comment = NULL,
  flatten = FALSE,
  defines = NULL,
  compression = NULL

  format = NULL,
  type = NULL,
  colorspace = NULL,
  depth = NULL,
  antialias = NULL,
  matte = NULL,
  interlace = NULL

image_data(image, channels = NULL, frame = 1)

image_raster(image, frame = 1, tidy = TRUE)

image_display(image, animate = TRUE)

image_browse(image, browser = getOption("browser"))


image_blank(width, height, color = "none", pseudo_image = "", defines = NULL)




image_get_artifact(image, artifact = "")




a file, url, or raster object or bitmap array


resolution to render pdf or svg


color depth (either 8 or 16)


drop image comments and metadata


automatically image_coalesce() gif images


a named character vector with extra options to control reading. These are the -define key{=value} settings in the command line tool. Use an empty string for value-less defines, and NA to unset a define.


in pixels


in pixels


integer vector with page numbers. Defaults to all pages.


user password to open protected pdf files


how many images to capture per second of video. Set to NULL to get all frames from the input video.


output format such as "png", "jpeg", "gif", "rgb" or "rgba".


magick image object returned by image_read() or image_graph()


number between 0 and 100 for jpeg quality. Defaults to 75.


text string added to the image metadata for supported formats


should image be flattened before writing? This also replaces transparency with background color.


a string with compression type from compress_types


string with imagetype value from image_types for example grayscale to convert into black/white


string with a colorspace from colorspace_types for example "gray", "rgb" or "cmyk"


enable anti-aliasing for text and strokes


set to TRUE or FALSE to enable or disable transparency


string with interlace


string with image channel(s) for example "rgb", "rgba", "cmyk","gray", or "ycbcr". Default is either "gray", "rgb" or "rgba" depending on the image


integer setting which frame to extract from the image


converts raster data to long form for use with geom_raster. If FALSE output is the same as as.raster().


support animations in the X11 display


argument passed to browseURL


a valid color string such as "navyblue" or "#000080". Use "none" for transparency.


string with pseudo image specification for example "radial-gradient:purple-yellow"


several images or lists of images to be combined


string with name of the artifact to extract, see the image_deskew for an example.


All standard base vector methods such as [, [[, c(), as.list(), as.raster(), rev(), length(), and print() can be used to work with magick image objects. Use the standard img[i] syntax to extract a subset of the frames from an image. The img[[i]] method is an alias for image_data() which extracts a single frame as a raw bitmap matrix with pixel values.

For reading svg or pdf it is recommended to use image_read_svg() and image_read_pdf() if the rsvg and pdftools R packages are available. These functions provide more rendering options (including rendering of literal svg) and better quality than built-in svg/pdf rendering delegates from imagemagick itself.

X11 is required for image_display() which is only works on some platforms. A more portable method is image_browse() which opens the image in a browser. RStudio has an embedded viewer that does this automatically which is quite nice.

Image objects are automatically released by the garbage collector when they are no longer reachable. Because the GC only runs once in a while, you can also call image_destroy() explicitly to release the memory immediately. This is usually only needed if you create a lot of images in a short period of time, and you might run out of memory.

See also


# Download image from the web frink <- image_read("") worldcup_frink <- image_fill(frink, "orange", "+100+200", 20) image_write(worldcup_frink, "output.png") # extract raw bitmap array bitmap <- frink[[1]] # replace pixels with #FF69B4 ('hot pink') and convert back to image bitmap[,50:100, 50:100] <- as.raw(c(0xff, 0x69, 0xb4, 0xff)) image_read(bitmap)
#> # A tibble: 1 × 7 #> format width height colorspace matte filesize density #> <chr> <int> <int> <chr> <lgl> <int> <chr> #> 1 PNG 220 445 sRGB TRUE 0 72x72
# Plot to graphics device via legacy raster format raster <- as.raster(frink) par(ask=FALSE) plot(raster)
# Read bitmap arrays from from other image packages curl::curl_download("", "example.webp") if(require(webp)) image_read(webp::read_webp("example.webp"))
#> Loading required package: webp
#> # A tibble: 1 × 7 #> format width height colorspace matte filesize density #> <chr> <int> <int> <chr> <lgl> <int> <chr> #> 1 PNG 550 404 sRGB TRUE 0 72x72
unlink(c("example.webp", "output.png")) if(require(rsvg)){ tiger <- image_read_svg("") svgtxt <- '<?xml version="1.0" encoding="UTF-8"?> <svg width="400" height="400" viewBox="0 0 400 400" fill="none"> <circle fill="steelblue" cx="200" cy="200" r="100" /> <circle fill="yellow" cx="200" cy="200" r="90" /> </svg>' circles <- image_read_svg(svgtxt) }
#> Loading required package: rsvg
#> Linking to librsvg 2.48.9
if(require(pdftools)) image_read_pdf(file.path(R.home('doc'), 'NEWS.pdf'), pages = 1, density = 100)
#> Loading required package: pdftools
#> Using poppler version 0.86.1
#> # A tibble: 1 × 7 #> format width height colorspace matte filesize density #> <chr> <int> <int> <chr> <lgl> <int> <chr> #> 1 PNG 850 1100 sRGB TRUE 0 100x100
# create a solid canvas image_blank(600, 400, "green")
#> # A tibble: 1 × 7 #> format width height colorspace matte filesize density #> <chr> <int> <int> <chr> <lgl> <int> <chr> #> 1 png 600 400 sRGB FALSE 0 72x72
image_blank(600, 400, pseudo_image = "radial-gradient:purple-yellow")
#> # A tibble: 1 × 7 #> format width height colorspace matte filesize density #> <chr> <int> <int> <chr> <lgl> <int> <chr> #> 1 RADIAL-GRADIENT 600 400 sRGB FALSE 0 72x72
image_blank(200, 200, pseudo_image = "gradient:#3498db-#db3a34", defines = c('gradient:direction' = 'east'))
#> # A tibble: 1 × 7 #> format width height colorspace matte filesize density #> <chr> <int> <int> <chr> <lgl> <int> <chr> #> 1 GRADIENT 200 200 sRGB FALSE 0 72x72