perhaps instead of stopifnot(inherits(x, "magick-image")) which gives Error: inherits(image, "magick-image") is not TRUE which I imagine some new to R may not understand

if (!inherits(x, "magick-image")) {
  stop('expecting object of class magick-image, class ', class(x), ' not supported')
}

function for converting a loaded image to bw/grayscale,

thanks for great package btw!

I was wondering how I would go about creating a GIF from multiple r plots with magick.

What i normally do is:

png(file="temp/example%02d.png")

for (i in subsets) {

    plot(subset(data, some.subset == i))

}

dev.off()

and then use the png in temp to create a GIF. I think magick can make this process easier but I do not really see how to go about.

i.e. how can I read just the first page of a pdf? With convert I can do path.pdf[0]

Similar to #42
http://www.imagemagick.org/script/command-line-options.php?#shear
http://www.imagemagick.org/Usage/warping/#shear

Image magick seems to have some ability to generate image metrics, as discussed here:
http://www.imagemagick.org/Usage/compare/#metrics

Ultimately, I hope to find ways to cluster images together and this might be helpful. I have found image_compare to be super useful so far in my work.

Related to my apparently very confusing question on the community call.

Various operations require user to specify a location or region in the Magick geometry specification, which is by default in pixels. This leads to lots of annoying arithmetic and trial and error for the user.

I was asking for ways to make this easier, such as:

1. specify a location relative to the image, e.g, seed the fill at c(0.5, 0.5) = middle of the image
2. overlay a grid or some sentinel points with pixel location indicated (like a football field)

As for point 1, it looks like the % geometry qualifier is very handy, so maybe worth featuring in docs or examples. Not sure if there's still room for magick to alleviate annoying arithmetic problems.

Enhance the preview with better image viewing. Some examples:

• jcrop widget to select a point or area.
• cropper.js
• cornerstone is a JS DICOM viewer that uses html5 canvas.
• DWV dicom viewer

I'm trying to batch process images to pad their canvas sizes so they fit a 3 x 2 ratio. This works in the command line with ImageMagick

convert mypic.jpg -background white -gravity center  -extent 1611x1074  mypic_3x2.jpg

As far as I can tell, magick implements the -background functionality with image_background(), but not gravity or extent. If at some point those could be added, that would be great. (Meanwhile, off to figure out how to run that command as a shell script from R....)

Thanks for the great package.

Dimensions don't seem to match bitmap size:

tiger <- image_read('https://upload.wikimedia.org/wikipedia/commons/f/fd/Ghostscript_Tiger.svg')
as.raster(tiger)

Setting line join doesn't work in combination with setting a linecap. This looks like a bug in IM.

# line joins
img <- image_graph()
par(ljoin=0)
y=c(0,1,0,5,0)
plot(y,lwd=20,type="l",ylim=c(0,10))
par(ljoin=1)
lines(y+2,lwd=20,type="l")
par(ljoin=2)
lines(y+4,lwd=20,type="l")
dev.off()

# mitre (doesn't work)
img <- image_graph()
par(ljoin=1) # lmitre only active for ljoin=1
y=c(0,30,0)
x=c(-1:1)
plot(x, y,lwd=10,type="l",ylim=c(0,40),xlim=c(-20,20))
par(ljoin=1, lmitre=30) # default lmitre=10
lines(x+4,y,lwd=10,type="l",ylim=c(0,40))
dev.off()

is there a way to suppress the print call at the end of the function?

if not can it be removed or add for a flag to control its print?

it is causing issues when we want to use the print function and are getting output in the console

magick:::`print.magick-image`
function (x, ...) 
{
   viewer <- getOption("viewer")
   ext <- ifelse(length(x), tolower(image_info(x[1])$format), 
       "gif")
   if (is.function(viewer)) {
       tmp <- file.path(tempdir(), paste0("preview.", ext))
       image_write(x, path = tmp)
       viewer(tmp)
   }
   print(image_info(x))
   invisible()
}

It would be great to use the phash algorithm for testing images.

Question. Why do you divide in as.double.bitmap by 255 and as.integer.bitmap the division is not done?

PDFs appear to be down sampled quite badly when you read them in currently. Is there a way to add a density argument to specify the resolution of a PDF when using image_read()? Something analogous to command line statements like convert -density 300 file.pdf file.png

I just captured some hand-drawings with a black "pen" then realized I needed to use them on a dark background. So I wanted to switch every pixel that is, say, "#FFFFFF" to "#000000".

I used the command line. This, cribbed from here, worked on the first try 😮:

mogrify -path ./ -format pdf -fill "#FFFFFF" -opaque "#000000" *.pdf

I didn't try it with magick yet, but it would be interesting to see if this is easy. If not, could it be? If it is, it might make a useful example.

Representative input file.
Arrows-jenny-2016-11-06.pdf

I'm trying to use image_montage perhaps I don't know how, but I get:

> image_read(files) %>% image_montage()
Fontconfig warning: line 146: blank doesn't take any effect anymore. please remove it from your fonts.conf
Fontconfig warning: line 146: blank doesn't take any effect anymore. please remove it from your fonts.conf
Error in eval(substitute(expr), envir, enclos) : 
  Magick: NoDecodeDelegateForThisImageFormat `' @ error/blob.c/ImagesToBlob/1813
> 
> traceback()
6: stop(list(message = "Magick: NoDecodeDelegateForThisImageFormat `' @ error/blob.c/ImagesToBlob/1813", 
       call = eval(substitute(expr), envir, enclos), cppstack = NULL))
5: .Call("magick_magick_image_write", PACKAGE = "magick", input, 
       format, quality)
4: magick_image_write(image, format, quality)
3: image_write(x, path = tmp)
2: `print.magick-image`(x)
1: function (x, ...) 
   UseMethod("print")(x)

This however seems to create something that makes sense:

> montage <- image_read(files) %>% image_montage()
> str( montage )
Class 'magick-image' <externalptr> 
> montage[[1]]
4 channel 768x504 bitmap array: raw [1:4, 1:768, 1:504] ff ff ff ff ...
 - attr(*, "class")= chr [1:2] "bitmap" "reba"

but then:

> montage
Error in eval(substitute(expr), envir, enclos) : 
  Magick: NoDecodeDelegateForThisImageFormat `' @ error/blob.c/ImagesToBlob/1813

This seems to work:

image_write(montage, format = "jpeg", path = "/tmp/test.jpeg")

Also, it seems that the image_montage function does not offer the geometry (or any other option actually, e.g. background, etc ...) control as in the underlying api

files that I use above is the list of these images

# git clone https://github.com/tmm1/emoji-extractor.git
files <- list.files( "emoji-extractor/images/160x160/", full.names = TRUE )

Are these images different?

library("magick")
logo <- image_read("https://www.r-project.org/logo/Rlogo.png")
image_border(logo, "red", "10x10")
image_border(logo, "red", "10x10+5-5")

If not, it seems like a bug.

See e.g. help('raster-package', package = 'raster')

Work around browser caching preview.png.

I get an error in a 1gb ubuntu machine at AWS:

Warning: Error in magick_image_write: Magick: memory allocation failed `' @ error/gif.c/WriteGIFImage/1648

The server used to work fine with 0.4 version, but today I updated the system and R packages and image_write does not work anymore. I've tried rolling back to 0.4 and I get the same error, so I guess it is a problem with libmagick++-dev which probably has been updated too.

Any hint on how to use magick in low RAM machines?

sessionInfo()

sessionInfo()
R version 3.4.1 (2017-06-30)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.3 LTS

Matrix products: default
BLAS: /usr/lib/libblas/libblas.so.3.6.0
LAPACK: /usr/lib/lapack/liblapack.so.3.6.0

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] memapp_2.4.20170802 magick_0.5

loaded via a namespace (and not attached):
[1] shiny_1.0.3 compiler_3.4.1 R6_2.2.2 htmltools_0.3.6
[5] Rcpp_0.12.12 digest_0.6.12 xtable_1.8-2 httpuv_1.3.5
[9] mime_0.5

Separate package to define bitmap classes and allow for conversion between color spaces (RGB, CMYK, etc) and various R raster formats (doubles, strings, etc).

Also take care of various ordering conventions. Maybe use boost gil for conversion.

I'm trying to plot an image directly to magick memory rather than round-tripping to disk, and can do so like so:

library(Cairo)
library(ggplot2)
library(magick)
x = 1680
y= 1050

Cairo(type="raster", width=x, height=y)
ggplot(mtcars, aes(x=cyl, y=mpg)) + geom_point()
i <- image_read(structure(packBits(intToBits(dev.capture(native=TRUE))), dim=c(4, x, y)))
dev.off()
image_write(i, "testmem.gif")

However, this is still slower than disk round-trip, with most of the time being taken up by translating the nativeRaster (8-bit ARGB, stored as integers) produced by dev.capture() to the right raw format:

It would be great if image_read() could also take nativeRasters for this use case and do the right conversion in a performant way. I think there are lots of applications where you'd like to manipulate plots without going to disk for good performance. (I'm currently trying to do this to update the gganimate package).

Wrapper for extracting text from images.

I am enjoying this package. However, I cannot figure out how to use image_append to create a thumbnail gallery of images. I have a table with the names of each png that needs to be added for a given week. I am trying to write something that will create 4 images across and have a variable number of rows, some weeks it will be 23 (6 rows) images, other weeks 75+ (19 rows). I've been trying to create a for loop that appends left_to_right and then top_to_bottom as shown in the vignette. Using assign(x, image_read) goes through but then says it isn't a magick image when I try to call it with get(). Any help would be appreciated.

Hello,

I'm trying to install magick 0.4 because it is a dependency of pkgdown. I've found that I cannot build it using gcc-7 on OSX (10.12.5), for R 3.4.0; all via homebrew. Switching to clang works though.

With gcc it compiles without issue but does not load:

** testing if installed package can be loaded
Error: package or namespace load failed for ‘magick’ in dyn.load(file, DLLpath = DLLpath, ...):
 unable to load shared object '/usr/local/lib/R/3.4/site-library/magick/libs/magick.so':
  dlopen(/usr/local/lib/R/3.4/site-library/magick/libs/magick.so, 6): Symbol not found: __ZN6Magick10labelImageC1ERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE
  Referenced from: /usr/local/lib/R/3.4/site-library/magick/libs/magick.so
  Expected in: flat namespace
 in /usr/local/lib/R/3.4/site-library/magick/libs/magick.so
Error: loading failed
Execution halted
ERROR: loading failed

Example gcc line:

g++-7 -std=gnu++11 -dynamiclib -Wl,-headerpad_max_install_names -undefined dynamic_lookup -single_module -multiply_defined suppress -L/usr/local/opt/openblas/lib -L/usr/local/opt/gettext/lib -L/usr/local/opt/readline/lib -L/usr/local/lib -L/usr/local/Cellar/r/3.4.0_1/R.framework/Resources/lib -L/usr/local/opt/openblas/lib -L/usr/local/opt/gettext/lib -L/usr/local/opt/readline/lib -L/usr/local/lib -o magick.so RcppExports.o attributes.o base.o config.o edit.o register.o transformations.o -L/usr/local/Cellar/imagemagick/7.0.5-9/lib -lMagick++-7.Q16HDRI -lMagickWand-7.Q16HDRI -lMagickCore-7.Q16HDRI -F/usr/local/Cellar/r/3.4.0_1/R.framework/.. -framework R -lintl -Wl,-framework -Wl,CoreFoundation

Here are the pkg-config notes:

Found pkg-config cflags and libs!
Using PKG_CFLAGS=-DMAGICKCORE_HDRI_ENABLE=1 -DMAGICKCORE_QUANTUM_DEPTH=16 -DMAGICKCORE_HDRI_ENABLE=1 -DMAGICKCORE_QUANTUM_DEPTH=16 -DMAGICKCORE_HDRI_ENABLE=1 -DMAGICKCORE_QUANTUM_DEPTH=16 -I/usr/local/Cellar/imagemagick/7.0.5-9/include/ImageMagick-7
Using PKG_LIBS=-L/usr/local/Cellar/imagemagick/7.0.5-9/lib -lMagick++-7.Q16HDRI -lMagickWand-7.Q16HDRI -lMagickCore-7.Q16HDRI

If I switch to clang it does compile successfully.

Example clang line:

clang++ -std=gnu++11 -dynamiclib -Wl,-headerpad_max_install_names -undefined dynamic_lookup -single_module -multiply_defined suppress -L/usr/local/opt/openblas/lib -L/usr/local/opt/gettext/lib -L/usr/local/opt/readline/lib -L/usr/local/lib -L/usr/local/Cellar/r/3.4.0_1/R.framework/Resources/lib -L/usr/local/opt/openblas/lib -L/usr/local/opt/gettext/lib -L/usr/local/opt/readline/lib -L/usr/local/lib -o magick.so RcppExports.o attributes.o base.o config.o edit.o register.o transformations.o -L/usr/local/Cellar/imagemagick/7.0.5-9/lib -lMagick++-7.Q16HDRI -lMagickWand-7.Q16HDRI -lMagickCore-7.Q16HDRI -F/usr/local/Cellar/r/3.4.0_1/R.framework/.. -framework R -lintl -Wl,-framework -Wl,CoreFoundation

Thanks,
Chris

Mac OSX

Using the OSX magick CRAN build, the magick graphics device has a few sec overhead on the first time of using fonts after a reboot:

fig <- image_device(res = 96)
ggplot2::qplot(mpg, wt, data = mtcars, colour = cyl)
dev.off()
print(fig)

After the first time it is gone. Even if we restart R. Perhaps fontconfig is rebuilding some tmp cache?

Windows

On Windows all font operations (both calc metrics and drawing) are consistently quite slow. Need to figure out how to debug what it is doing.

FontCondfig Debugging

The fontoconfig manual pages:

Name         Value    Meaning
  ---------------------------------------------------------
  MATCH            1    Brief information about font matching
  MATCHV           2    Extensive font matching information
  EDIT             4    Monitor match/test/edit execution
  FONTSET          8    Track loading of font information at startup
  CACHE           16    Watch cache files being written
  CACHEV          32    Extensive cache file writing information
  PARSE           64    (no longer in use)
  SCAN           128    Watch font files being scanned to build caches
  SCANV          256    Verbose font file scanning information
  MEMORY         512    Monitor fontconfig memory usage
  CONFIG        1024    Monitor which config files are loaded
  LANGSET       2048    Dump char sets used to construct lang values
  MATCH2        4096    Display font-matching transformation in patterns

So we can get some debugging info:

Sys.setenv(FC_DEBUG = 1 + 4 + 8 + 16 + 128 + 1024)
library(magick)
fig <- image_device(res = 96)
ggplot2::qplot(mpg, wt, data = mtcars, colour = cyl)
dev.off()
print(fig)

Any tips would be amazing :) @lionel- @jimhester @hadley

Can appear on macOS when you have an old version of fontconfig. Upgrading XQuartz to the latest XQuartz-2.7.10 resolves the issue: https://www.xquartz.org/releases/index.html.

Upstream issue: bug 97493.

Hi there,
I tried to follow the ropensci tutorial about magick, and I found an error. I'm using R 3.3.2 (also tried with previous version) on windows 10, 64 bit.

in

# foreground
banana <- image_read(system.file("banana.gif", package = "magick"))

my r process doesn't stop.
I have to terminate it.

So I broke down the problem:

• When I search for the file with only system.file("banana.gif", package = "magick") it returns ""
• when I do a image_read("") the r process hangs (?) or at least does not terminate or error. it just keeps going.

Hello!
This is a request for a function that returns a data representation of a color histogram on a per-image bases and or return a list/vector for a set of images.

We are doing research where we are looking for differences across sets of images.
For example, we might like to find the average histogram of image set A. Next we would find it for image set B and then see if we can find the distance between A and B.

Is this possible?

Similar to #15, The [[ function's code should be changed from image_write to image_write_bitmap.

The image_raster callback currently rotates around it's centerpoint but the cairo device rotates around (x,y):

# use standard device
graphics.off()
example(rasterImage)

# using magick
img <- image_graph()
example(rasterImage)

To do the same we need to calculate the offset in the lower left corner. Something like this?

rot <- (rot + 360) %% 360;
rad <- (rot * pi) / 180
x_offset <- width * cos(rad) + height * sin(rad)
y_offset <- height * cos(rad) +  width * sin(rad)

God I'm getting old.

https://www.imagemagick.org/script/command-line-options.php#deskew

I could write my img to file and wrap a system command, then read the result back into R, but that would be really slow...

Could you suggest a method of calling imagemagick with -deskew?

I think we need to implement S4 methods for:

raster  To create a RasterLayer
stack   To create a RasterStack (multiple layers)
brick   To create a RasterBrick (multiple layers)

A polyline from ggplot2 seems to close.

update: this happens only for a certain DrawableStrokeLineJoin which ggplot2 uses.

library(ggplot2)
img <- magick::image_device()
qplot(cars$speed, cars$dist, geom = c("smooth"), se = FALSE) + theme_void()
dev.off()
img

But this is fine:

img <- image_device(bg = 'white');
plot.new(); 
plot.window(c(0,1), c(0,1))
lines(c(0.1, 0.3, 0.2), c(0.1, 0.5, 0.8), col = "blue", lwd = 3)
dev.off()
img

Together:

library(ggplot2)
library(magick)
dev <- magick::image_device()
plot.new()
qplot(cars$speed, cars$dist, geom = c("smooth"), se = FALSE) + theme_void()
lines(c(0.1, 0.3, 0.2), c(0.1, 0.5, 0.8), col = "red", lwd = 3)
dev.off()
dev[2]

i don't' know how common this would be, but perhaps users may want to import image from a location that requires auth or proxy, etc.

Would be great. :)

The coposite operator affects the original image, even though we set it on a copy:

library(magick)

bigdata <- image_read('http://feelgrafix.com/data_images/out/28/1004158-data.jpg')
frink <- image_read("https://jeroenooms.github.io/images/frink.png")
logo <- image_read("https://www.r-project.org/logo/Rlogo.png")
img <- image_scale(c(bigdata, logo, frink), "400x400")

image_flatten(img)
image_flatten(img, "Add")
image_flatten(img)

Currently we simply call the rstudio viewer function to preview an image. Perhaps we can instead wrap it into an html widget so that the same figure will automatically be inserted into the document by knitr or shiny?

Not sure if this makes any sense at all, hopefully @timelyportfolio or @jcheng5 can chime in.

I'm using -background white -alpha remove to replace transparent pixels with white. I can't see how to replicate that with magick.

I am using magick to prepare images for tesseract's ocr() function (rOpenSci package) and with some images I get the following error when I also used image_deskew(): Magick: TIFF: negative image positions unsupported.

The possible solution (using repage) seems to not be implemented?

Thank you!

The use case I have in mind is to clean-up images before feeding them to tesseract. I personally have seen great improvements running lat on images before feeding to tesseractbut magick has no bindings for this? I can't seem to achieve the same effect with a combination of image_contrast, image_blur, and image_enhance.

Include R_ext/Visibility.h to control hide unexported stuff from packages.

See https://www.imagemagick.org/Magick++/Drawable.html

i know it'd be easy to just png::readPNG() but any thoughts on an as_raser() capability to get this into a type other img pkgs already handle? (i can poke at this as well in a few weeks)

** package ‘magick’ successfully unpacked and MD5 sums checked
Found pkg-config cflags and libs!
Using PKG_CFLAGS=-fopenmp -DMAGICKCORE_HDRI_ENABLE=1 -DMAGICKCORE_QUANTUM_DEPTH=16 -fopenmp -DMAGICKCORE_HDRI_ENABLE=1 -DMAGICKCORE_QUANTUM_DEPTH=16 -fopenmp -DMAGICKCORE_HDRI_ENABLE=1 -DMAGICKCORE_QUANTUM_DEPTH=16 -I/usr/local/Cellar/imagemagick/7.0.5-4/include/ImageMagick-7
Using PKG_LIBS=-L/usr/local/Cellar/imagemagick/7.0.5-4/lib -lMagick++-7.Q16HDRI -lMagickWand-7.Q16HDRI -lMagickCore-7.Q16HDRI
------------------------- ANTICONF ERROR ---------------------------
Configuration failed because Magick++ was not found. Try installing:
 * deb: libmagick++-dev (Debian, Ubuntu)
 * rpm: ImageMagick-c++-devel (Fedora, CentOS, RHEL)
 * csw: imagemagick_dev (Solaris)
 * brew: imagemagick (OSX)
If Magick++ is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a Magick++.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'

brew info

/usr/local/Cellar/imagemagick/7.0.5-4 (1,524 files, 23MB) *
  Built from source on 2017-03-26 at 18:14:18 with: --with-openmp --with-fontconfig --with-librsvg --with-ghostscript --with-webp --with-pango

I assume one of the options I built imagemagick with is causing problems? I will rebuild with your README suggestion and report back.

Here is an example: http://lokeshdhakar.com/projects/color-thief/

I don't know if this kind of functionality is implemented in imagemagick, but finding and comparing palettes would be quite useful.

I just discovered that I had over 100 GB of magick temp files. I believe they were left when an rmagick function locked up and I had to force quit R. Some way of making sure these get cleaned would be a bonus.

Add a link to https://vimeo.com/channels/rocommunitycalls/180799058 in the README?

@stefaniebutland reminded me it existed.

to make ?magick possible