diff --git a/documentation/Index.md b/documentation/Index.md
index c98e248efef019be1db7917eab74484fd939a41a..393dcd24acb72e76d2e3f3c05b408a00b2986d3d 100644
--- a/documentation/Index.md
+++ b/documentation/Index.md
@@ -7,3 +7,5 @@
 [For Developers](dev/Index.md)
 
 [Debugging](debugging.md)
+
+[Graphics support](graphics.md)
diff --git a/documentation/graphics.md b/documentation/graphics.md
new file mode 100644
index 0000000000000000000000000000000000000000..e5b13a857ae5abe437e365b04e3f3ef4a2ce4d94
--- /dev/null
+++ b/documentation/graphics.md
@@ -0,0 +1,88 @@
+# Introduction
+
+There are two main built-in R packages that provide basic graphical output 
+support: *graphics* and *lattice*. They both use some parts of even lower level 
+*grDevices* package, which allows users to change the "device" to which the 
+graphical output will be drawn. The probably most popular graphical packages 
+*lattice* and *ggplot2* are build on top of the *grid* package.
+
+FastR has its own implementation of the *grid* package and it also emulates 
+the most important R level functions from the *grDevices* package. This 
+implementation is purely Java based and has been tested to work with 
+the unmodified *lattice* package, support for unmodified *ggplot2* package 
+is work in progress.
+
+Any functions from the *graphics* package are not going to work with FastR, 
+namely e.g. the `plot` function. R documentation, which can be displayed by 
+typing `?functionName` can be used to determine if a function is from *graphics* 
+or *grid* package or potentially from some *grid* based higher level package.
+
+The longer term goal is to emulate R level *graphics* package functions by 
+means of *grid* package drawing primitives.
+
+# Examples
+
+The *grid* package is distributed with FastR, but not loaded by default, 
+therefore one has to load it using the `library` function. Following example 
+draws a red rectangle in the center of the screen.
+
+```
+library(grid)
+grid.rect(width=0.5, height=0.5, gp=gpar(col='red'))
+```
+
+The *lattice* package must be installed using:
+
+```
+install.packages('lattice');
+```
+
+With the *lattice* package installed, one can run the 
+following example showing a barchart.
+
+```
+library(lattice)
+mtcars$cars <- rownames(mtcars)
+barchart(cars~mpg, data=mtcars)
+```
+
+# Devices Support
+
+FastR supports the following devices:
+
+* awt: opens a new AWT window for drawing, the window can be resized and 
+its contents can be saved with the standard `savePlot` function. 
+This is the default device for interactive sessions. 
+In FastR, for compatibility with GNU R, *X11* device is alias for the *awt* device.
+
+* svg: generates SVG image, which is saved into a specified file. 
+The contents may be read into a character vector by FastR specific function 
+`grDevices:::svg.off()`. FastR produces the SVG code directly from *grid* drawing commands, 
+therefore the SVG code may look more SVG idiomatic and resembling the R code that created it, 
+but may produce visually slightly different image compared to other devices. 
+This device is default for non-interactive FastR sessions.
+
+* jpg, png, bmp: these devices create an image file of specified format. Unlike the SVG device, 
+these are also based on the AWT framework.
+
+Devices that generate file take filename as an argument. The file will be saved on disk only 
+once the standard `dev.off()` function is called. FastR honors the "device" option, which 
+may override the default device. See the documentation of *grDevices* package for more details.
+
+# Java Interoperability
+
+The `grDevices::awt` function can accept `java.awt.Graphics2D` object and width and height in AWT units, 
+in which case all the drawing will be done using that object. Example:
+
+```
+grDevices:::awt(420, 420, graphicsObj);
+```
+
+One possible use case is to create a Java based UI with a `JPanel` that will be 
+used for visualizing some data from R. Override the `JPanel`'s `paint` 
+method and pass the graphics object to R code using `PolyglotEngine`. 
+The R code can do any *grid* based visualization and it will be directly 
+displayed in the UI.
+
+
+