@mainpage libdcp
-libdcp is a small library to create Digital Cinema Packages (DCPs) from JPEG2000 and WAV files.
+libdcp is a small library to create and read Digital Cinema Packages (DCPs) from JPEG2000 and WAV files.
Most of the hard work is done by a (slightly patched) version of asdcplib (http://www.cinecert.com/asdcplib/)
which is included in the source distribution for libdcp.
-See libdcp::DCP for an example of libdcp's use.
-
libdcp is distributed under the GNU GPL.
+Creating DCPs
+--
+
+Typical use for creating DCPs might be:
+@code
+#include <libdcp/dcp.h>
+using namespace std;
+
+libdcp::DCP dcp ("My Film DCP", "My Film", libdcp::DCP::FEATURE, 24, 50000);
+
+vector<string> j2k_files;
+j2k_files.push_back ("1.j2c");
+...
+j2k_files.push_back ("50000.j2c");
+
+// These images are 1998x1080 pixels (DCI Flat)
+dcp.add_picture_asset (j2k_files, 1998, 1080);
+
+vector<string> wav_files;
+wav_files.push_back ("L.wav");
+wav_files.push_back ("R.wav");
+wav_files.push_back ("C.wav");
+wav_files.push_back ("Lfe.wav");
+wav_files.push_back ("Ls.wav");
+wav_files.push_back ("Rs.wav");
+dcp.add_sound_asset (wav_files);
+
+dcp.write_xml ();
+
+@endcode
+
+This will create a DCP at 24 frames per second with 50000 frames, writing
+data to a directory "My Film DCP", naming the DCP "My Film" and marking
+as a Feature. We then add the picture and sound files (which creates
+MXF files inside the DCP directory) and then write the required XML files.
+
+If you want to report progress for long jobs (add_picture_asset() can
+take a long time, in particular, as it must do a lot of disk I/O for
+large DCPs), connect to the libdcp::DCP::Progress signal and report its parameter
+to the user (it will range between 0 for 0% and 1 for 100%) using something like
+
+@code
+void
+report (float p)
+{
+ cout << "We are " << int (p * 100) << "% done.\n";
+}
+
+dcp.Progress.connect (sigc::ptr_fun (&report));
+@endcode
+
+If you can generate content paths algorithmically, you may prefer to do something
+like this:
+
+@code
+
+string
+j2k_path (int frame)
+{
+ stringstream s;
+ s << "my_j2ks/" << frame << ".j2c"
+ return s.str ();
+}
+
+string
+wav_path (libdcp::Channel channel)
+{
+ switch (channel) {
+ case LEFT:
+ return "left.wav";
+ case RIGHT:
+ return "right.wav";
+ }
+
+ return "";
+}
+
+...
+
+// Our images are 1998x1080 (DCI Flat)
+dcp.add_picture_asset (sigc::ptr_fun (&j2k_path), 1998, 1080);
+// We have two sound channels
+dcp.add_sound_asset (sigc::ptr_fun (&wav_path), 2);
+
+...
+
+@endcode
+
+Reading existing DCPs
+--
+
+Alternatively libdcp can be used to read existing DCPs and examine their content. You
+might do
+
+@code
+#include <libdcp/dcp.h>
+using namespace std;
+
+libdcp::DCP dcp ("some-DCP-directory");
+@endcode
+
+libdcp will look at the XML files that make up the DCP and find its assets. You can then
+do things like
+
+@code
+boost::shared_ptr<libdcp::PictureAsset> p = dcp.picture_asset ();
+boost::shared_ptr<libdcp::RGBAFrame> f = p->get_frame(42)->rgba_frame ();
+uint8_t* data = f->data ();
+int size = f->size ();
+@endcode
+
+This will extract the image of frame 42 from the DCP and make its RGBA data available
+for examination.
+
+Audio data is accessed in chunks equal in length to the duration of a video frame. To
+get the audio that accompanies frame 66, you can do
+
+@code
+boost::shared_ptr<libdcp::SoundAsset> s = dcp.sound_asset ();
+cout << "Sound has " << s->channels() << " channels at " << s->sampling_rate() << "Hz\n";
+boost::shared_ptr<libdcp::SoundFrame> f = s->get_frame(66);
+uint8_t* data = f->data ();
+int size = f->size ();
+@endcode
+
+The returned data is interleaved 24-bit samples, so that
+
+@code
+int left = data[0] | (data[1] << 8) | (data[2] << 16);
+data += 3;
+int right = data[0] | (data[1] << 8) | (data[2] << 16);
+data += 3;
+int centre = data[0] | (data[1] << 8) | (data[2] << 16);
+data += 3;
+int lfe = data[0] | (data[1] << 8) | (data[2] << 16);
+data += 3;
+int ls = data[0] | (data[1] << 8) | (data[2] << 16);
+data += 3;
+int rs = data[0] | (data[1] << 8) | (data[2] << 16);
+data += 3;
+@endcode
+
+would obtain the first sample from each of the 6 channels for that frame.
+
+Subtitles can be read using code like
+
+@code
+boost::shared_ptr<SubtitleAsset> s = dcp.subtitle_asset ();
+list<boost::shared_ptr<libdcp::Text> > texts = s->subtitles_at (libdcp::Time (0, 3, 2, 5));
+@endcode
+
+This returns the subtitles that should be shown at 3 minutes, 2
+seconds, 5 ticks (where 1 tick is 4ms) into the DCP.
+
*/
projects / libdcp.git / blobdiff