+included in the image and not overlaid by the projector) or included
+as a separate subtitle ‘asset’ within your DCP (in which
+case the projector overlays them onto the image on playback). The
+difference between these two arrangements is illustrated by <xref
+linkend="fig-burn-in"/> and <xref linkend="fig-discrete"/>
+</para>
+
+<figure id="fig-burn-in">
+ <title>Burnt-in subtitles</title>
+ <mediaobject>
+ <imageobject>
+ <imagedata scale="80" fileref="diagrams/burn-in&dia;"/>
+ </imageobject>
+ </mediaobject>
+</figure>
+
+<figure id="fig-discrete">
+ <title>Separate subtitles</title>
+ <mediaobject>
+ <imageobject>
+ <imagedata scale="80" fileref="diagrams/discrete&dia;"/>
+ </imageobject>
+ </mediaobject>
+</figure>
+
+<para>
+The advantage of separate subtitles is that the same video content can
+be used for DCPs in many different languages. This means that only a
+small text file needs to be changed for each target language, rather
+than a large video file. It also means that the time-consuming video
+encoding need only be done once for the project rather than once for
+every language.
+</para>
+
+<para>
+Note that subtitles come in two types: text and bitmap. Text
+subtitles are expressed as plain text and can be either burnt into the
+image or included as a separate subtitle asset within the DCP. Bitmap
+subtitles, on the other hand, are expressed as pre-rendered bitmaps.
+They cannot (yet) be added to the DCP as a separate asset and must be
+burnt into the image.