diff --git a/Makefile b/Makefile
index 6efa8b7..5c189d9 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 clean:
-	rm -r out/
+	rm -r out/ out.epub
 
 example-gpub:
 	cd example-gpub; zip -r ../example-book.gpub *; cd -
diff --git a/README.md b/README.md
index 71952f1..30967ce 100644
--- a/README.md
+++ b/README.md
@@ -26,15 +26,36 @@ awk -f g2e.awk example-gpub/
 
 # notes
 
-the converter works with gempub archives that have a flat file structure: all the `.gmi` files, including `index.gmi`, have to be in the same directory. see the provided `example-gpub/`.
+## file structure
+
+the converter works with gempub archives that have a flat file structure: all the `.gmi` files, including `index.gmi`, have to be in the same directory.
+
+furthermore, the image assets for the book have to be inside that directory, within a directory called `img/`.
+
+for example, the structure for `example-gpub` is:
+
+```
+example-gpub/
+  metadata.txt
+  book/
+    index.gmi
+    cover.gmi
+    chapter1.gmi
+    chapter2.gmi
+    chapter3.gmi
+    img/
+      screencap_uxn-pong.gif
+      screenshot_uxn-pong-paddle.png
+      screenshot_uxn-pong-paddles-and-ball.png
+      screenshot_uxn-pong-paddles.png
+```
+
+note that only `.gmi`, `.png`, `.jpg` and `.gif` files are handled.
+
+## raw html
 
 you can insert raw html using the corresponding `.gmo` syntax: the contents of a line that starts with `+` will be inserted as is.
 
-# TODO
-
-* handle non-gmi files in the manifest
-* handle links to images
-
 # references
 
 * [EPUB - Wikipedia](https://en.wikipedia.org/wiki/EPUB#Implementation)
diff --git a/example-gpub/book/chapter1.gmi b/example-gpub/book/chapter1.gmi
index f6010bd..01403ba 100644
--- a/example-gpub/book/chapter1.gmi
+++ b/example-gpub/book/chapter1.gmi
@@ -2,6 +2,8 @@
 
 hello!
 
+=> img/screencap_uxn-pong.gif animated pong
+
 > not everything
 
 ## this is a second-level heading
diff --git a/example-gpub/book/cover.gmi b/example-gpub/book/cover.gmi
new file mode 100644
index 0000000..cf4445e
--- /dev/null
+++ b/example-gpub/book/cover.gmi
@@ -0,0 +1,5 @@
+# an example book
+
+a gpub archive that gets converted to an epub file
+
+=> img/screenshot_uxn-pong-paddles-and-ball.png screenshot of pong
diff --git a/example-gpub/book/img/screencap_uxn-pong.gif b/example-gpub/book/img/screencap_uxn-pong.gif
new file mode 100644
index 0000000..410f914
Binary files /dev/null and b/example-gpub/book/img/screencap_uxn-pong.gif differ
diff --git a/example-gpub/book/img/screenshot_uxn-pong-paddle.png b/example-gpub/book/img/screenshot_uxn-pong-paddle.png
new file mode 100644
index 0000000..d7b011e
Binary files /dev/null and b/example-gpub/book/img/screenshot_uxn-pong-paddle.png differ
diff --git a/example-gpub/book/img/screenshot_uxn-pong-paddles-and-ball.png b/example-gpub/book/img/screenshot_uxn-pong-paddles-and-ball.png
new file mode 100644
index 0000000..81257e5
Binary files /dev/null and b/example-gpub/book/img/screenshot_uxn-pong-paddles-and-ball.png differ
diff --git a/example-gpub/book/img/screenshot_uxn-pong-paddles.png b/example-gpub/book/img/screenshot_uxn-pong-paddles.png
new file mode 100644
index 0000000..717f80c
Binary files /dev/null and b/example-gpub/book/img/screenshot_uxn-pong-paddles.png differ
diff --git a/example-gpub/book/index.gmi b/example-gpub/book/index.gmi
index a0afd33..75ca361 100644
--- a/example-gpub/book/index.gmi
+++ b/example-gpub/book/index.gmi
@@ -2,6 +2,7 @@
 
 hello!
 
+=> cover.gmi an example book
 => chapter1.gmi chapter 1: an example
 => chapter2.gmi chapter 2: another example
 => chapter3.gmi chapter 3: yet another
diff --git a/g2e.awk b/g2e.awk
index 100262e..d98642a 100644
--- a/g2e.awk
+++ b/g2e.awk
@@ -35,10 +35,13 @@ BEGIN{
 
 	# setup epub
 	system( "mkdir -p " epubodir "META-INF" )
-	system( "mkdir -p " epubodir indexdirname )
+	system( "mkdir -p " epubodir indexdirname "img" )
 	system( "cp -u " tdir "container.xml " epubodir "META-INF/" )
 	system( "cp -u " tdir "mimetype " epubodir )
 	system( "cp -u " tdir "style.css " epubodir )
+	# copy images
+#	system( "cp -u " gpubdir indexdirname "img/* " epubodir indexdirname "img/" )
+
 
 	# read templates
 	templatefiles = "find " tdir " -type f -not -name '.*'"
@@ -94,6 +97,21 @@ ARGIND==1{ # skip other lines of the index
 
 # when finished reading the index:
 ARGIND==2 && FNR==1{
+	line = $0
+	# add images to manifest
+	while( "find " gpubdir indexdirname "img/ -type f -regextype awk -iregex \".+(png|jpg|gif)$\""  | getline){
+		path = $0
+		dest = path; sub(gpubdir, "", dest)
+		fortoc = dest
+		system( "cp -u " $0 " " epubodir dest) # copy images
+		mediatype = fortoc~/gif$/ ? "image/gif" : fortoc~/jpg$/ ? "image/jpeg" : "image/png"
+		id = fortoc
+		sub(/^.+\//,"",id)
+		content = content "    <item id=\"" id "\" href=\"" fortoc "\" media-type=\"" mediatype "\"/>\n"
+	}
+	$0 = line
+
+	# finalize metadata files
 	content = content "  </manifest>\n\n" spinetoc "  </spine>\n\n</package>"
 	printf content > epubodir "content.opf"
 
@@ -175,10 +193,18 @@ sub(/^>[[:space:]]*/,""){
 # links?
 sub(/^=>[[:space:]]*/,""){
 	link = $1
-	if(link~/gmi$/) sub(/gmi$/,"xhtml",link)
 	text = $2
 	for(i=3;i<=NF;i++) text = text " " $i
-	append("<p><a href=\"" link "\">" text "</a></p>")
+	if(link~/gmi$/){
+		sub(/gmi$/,"xhtml",link)
+		append("<p><a href=\"" link "\">" text "</a></p>")
+	}
+	else if(link~/(gif|jpg|png)$/){
+		append("<img src=\"" link "\" alt=\"" text "\" />")
+	}
+	else{
+		append("<p><a href=\"" link "\">" text "</a></p>")
+	}
 	next
 }