return mlt_deque_count( this->stack_image ) == 0 || mlt_properties_get_int( MLT_FRAME_PROPERTIES( this ), "test_image" );
}
-/** Check if we have a way to derive something than test audio.
+/** Check if we have a way to derive something other than test audio.
*/
int mlt_frame_is_test_audio( mlt_frame this )
{
- return this->get_audio == NULL || mlt_properties_get_int( MLT_FRAME_PROPERTIES( this ), "test_audio" );
+ return mlt_deque_count( this->stack_audio ) == 0 || mlt_properties_get_int( MLT_FRAME_PROPERTIES( this ), "test_audio" );
}
/** Get the aspect ratio of the frame.
return this->stack_service;
}
+/** [EXPERIMENTAL] Replace image stack with the information provided.
+
+ This might prove to be unreliable and restrictive - the idea is that a transition
+ which normally uses two images may decide to only use the b frame (ie: in the case
+ of a composite where the b frame completely obscures the a frame).
+
+ The image must be writable and the destructor for the image itself must be taken
+ care of on another frame and that frame cannot have a replace applied to it...
+ Further it assumes that no alpha mask is in use.
+
+ For these reasons, it can only be used in a specific situation - when you have
+ multiple tracks each with their own transition and these transitions are applied
+ in a strictly reversed order (ie: highest numbered [lowest track] is processed
+ first).
+
+ More reliable approach - the cases should be detected during the process phase
+ and the upper tracks should simply not be invited to stack...
+*/
+
+void mlt_frame_replace_image( mlt_frame this, uint8_t *image, mlt_image_format format, int width, int height )
+{
+ // Herein lies the potential problem for this function - it makes a potentially
+ // dangerous assumption that all content on the image stack can be removed without a destructor
+ while( mlt_deque_pop_back( this->stack_image ) ) ;
+
+ // Update the information
+ mlt_properties_set_data( MLT_FRAME_PROPERTIES( this ), "image", image, 0, NULL, NULL );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES( this ), "width", width );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES( this ), "height", height );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES( this ), "format", format );
+ this->get_alpha_mask = NULL;
+}
+
+/** Get the image associated to the frame.
+*/
+
int mlt_frame_get_image( mlt_frame this, uint8_t **buffer, mlt_image_format *format, int *width, int *height, int writable )
{
mlt_properties properties = MLT_FRAME_PROPERTIES( this );
int mlt_frame_get_audio( mlt_frame this, int16_t **buffer, mlt_audio_format *format, int *frequency, int *channels, int *samples )
{
+ mlt_get_audio get_audio = mlt_frame_pop_audio( this );
mlt_properties properties = MLT_FRAME_PROPERTIES( this );
int hide = mlt_properties_get_int( properties, "test_audio" );
- if ( hide == 0 && this->get_audio != NULL )
+ if ( hide == 0 && get_audio != NULL )
{
mlt_position position = mlt_frame_get_position( this );
- this->get_audio( this, buffer, format, frequency, channels, samples );
+ get_audio( this, buffer, format, frequency, channels, samples );
mlt_frame_set_position( this, position );
}
else if ( mlt_properties_get_data( properties, "audio", NULL ) )
return this->process( this, a_frame, b_frame );
}
-/** Get a frame from this filter.
-
- The logic is complex here. A transition is applied to frames on the a and b tracks
- specified in the connect method above. Since all frames are obtained via this
- method for all tracks, we have to take special care that we only obtain the a and
- b frames once - we do this on the first call to get a frame from either a or b.
-
- After that, we have 2 cases to resolve:
-
- 1) if the track is the a_track and we're in the time zone, then we need to call the
- process method to do the effect on the frame and remember we've passed it on
- otherwise, we pass on the a_frame unmolested;
- 2) For all other tracks, we get the frames on demand.
+/** Get a frame from this transition.
+
+ The logic is complex here. A transition is typically applied to frames on the a and
+ b tracks specified in the connect method above and only if both contain valid info
+ for the transition type (this is either audio or image).
+
+ However, the fixed a_track may not always contain data of the correct type, eg:
+
+ +---------+ +-------+
+ |c1 | |c5 | <-- A(0,1) <-- B(0,2) <-- get frame
+ +---------+ +---------+-+-----+ | |
+ |c4 | <------+ |
+ +----------+-----------+-+---------+ |
+ |c2 |c3 | <-----------------+
+ +----------+-------------+
+
+ During the overlap of c1 and c2, there is nothing for the A transition to do, so this
+ results in a no operation, but B is triggered. During the overlap of c2 and c3, again,
+ the A transition is inactive and because the B transition is pointing at track 0,
+ it too would be inactive. This isn't an ideal situation - it's better if the B
+ transition simply treats the frames from c3 as though they're the a track.
+
+ For this to work, we cache all frames coming from all tracks between the a and b
+ tracks. Before we process, we determine that the b frame contains someting of the
+ right type and then we determine which frame to use as the a frame (selecting a
+ matching frame from a_track to b_track - 1). If both frames contain data of the
+ correct type, we process the transition.
+
+ This method is invoked for each track and we return the cached frames as needed.
+ We clear the cache only when the requested frame is flagged as a 'last_track' frame.
*/
static int transition_get_frame( mlt_service service, mlt_frame_ptr frame, int index )
{
+ int error = 0;
mlt_transition this = service->child;
mlt_properties properties = MLT_TRANSITION_PROPERTIES( this );
- int accepts_blanks = mlt_properties_get_int( properties, "_accepts_blanks" );
+ int accepts_blanks = mlt_properties_get_int( properties, "accepts_blanks" );
int a_track = mlt_properties_get_int( properties, "a_track" );
int b_track = mlt_properties_get_int( properties, "b_track" );
mlt_position in = mlt_properties_get_position( properties, "in" );
mlt_position out = mlt_properties_get_position( properties, "out" );
int always_active = mlt_properties_get_int( properties, "always_active" );
+ int type = mlt_properties_get_int( properties, "_transition_type" );
+ int reverse_order = 0;
- if ( ( index == a_track || index == b_track ) && !( this->a_held || this->b_held ) )
+ // Ensure that we have the correct order
+ if ( a_track > b_track )
{
- mlt_service_get_frame( this->producer, &this->a_frame, a_track );
- mlt_service_get_frame( this->producer, &this->b_frame, b_track );
- this->a_held = 1;
- this->b_held = 1;
+ reverse_order = 1;
+ a_track = b_track;
+ b_track = mlt_properties_get_int( properties, "a_track" );
}
-
- // Special case track processing
- if ( index == a_track )
+
+ // Only act on this operation once per multitrack iteration from the tractor
+ if ( !this->held )
{
- // Determine if we're in the right time zone
- mlt_position position = mlt_frame_get_position( this->a_frame );
- if ( always_active || ( position >= in && position <= out ) )
+ int active = 0;
+ int i = 0;
+ int a_frame = a_track;
+ int b_frame = b_track;
+ mlt_position position;
+ int ( *invalid )( mlt_frame ) = type == 1 ? mlt_frame_is_test_card : mlt_frame_is_test_audio;
+
+ // Initialise temporary store
+ if ( this->frames == NULL )
+ this->frames = calloc( sizeof( mlt_frame ), b_track + 1 );
+
+ // Get all frames between a and b
+ for( i = a_track; i <= b_track; i ++ )
+ mlt_service_get_frame( this->producer, &this->frames[ i ], i );
+
+ // We're holding these frames until the last_track frame property is received
+ this->held = 1;
+
+ // When we need to locate the a_frame
+ switch( type )
{
- if ( !accepts_blanks && ( this->b_frame == NULL || ( mlt_frame_is_test_card( this->b_frame ) && mlt_frame_is_test_audio( this->b_frame ) ) ) )
- {
- *frame = this->a_frame;
- }
- else if ( !accepts_blanks && ( this->a_frame == NULL || ( mlt_frame_is_test_card( this->a_frame ) && mlt_frame_is_test_audio( this->a_frame ) ) ) )
- {
- mlt_frame t = this->a_frame;
- this->a_frame = this->b_frame;
- this->b_frame = t;
- *frame = this->a_frame;
- }
- else
- {
- int hide = 0;
- *frame = mlt_transition_process( this, this->a_frame, this->b_frame );
- if ( !mlt_properties_get_int( MLT_FRAME_PROPERTIES( this->a_frame ), "test_image" ) )
- hide = 1;
- if ( !mlt_properties_get_int( MLT_FRAME_PROPERTIES( this->a_frame ), "test_audio" ) )
- hide |= 2;
- mlt_properties_set_int( MLT_FRAME_PROPERTIES( this->b_frame ), "hide", hide );
- }
- this->a_held = 0;
+ case 1:
+ case 2:
+ // Some transitions (esp. audio) may accept blank frames
+ active = accepts_blanks;
+
+ // If we're not active then...
+ if ( !active )
+ {
+ // Hunt for the a_frame
+ while( a_frame <= b_frame && invalid( this->frames[ a_frame ] ) )
+ a_frame ++;
+
+ // Determine if we're active now
+ active = a_frame != b_frame && !invalid( this->frames[ b_frame ] );
+ }
+ break;
+
+ default:
+ fprintf( stderr, "invalid transition type\n" );
+ break;
}
- else
+
+ // Now handle the non-always active case
+ if ( active && !always_active )
{
- // Pass on the 'a frame' and remember that we've done it
- *frame = this->a_frame;
- this->a_held = 0;
+ // For non-always-active transitions, we need the current position of the a frame
+ position = mlt_frame_get_position( this->frames[ a_frame ] );
+
+ // If a is in range, we're active
+ active = position >= in && position <= out;
+ }
+
+ // Finally, process the a and b frames
+ if ( active )
+ {
+ mlt_frame a_frame_ptr = this->frames[ !reverse_order ? a_frame : b_frame ];
+ mlt_frame b_frame_ptr = this->frames[ !reverse_order ? b_frame : a_frame ];
+ int a_hide = mlt_properties_get_int( MLT_FRAME_PROPERTIES( a_frame_ptr ), "hide" );
+ int b_hide = mlt_properties_get_int( MLT_FRAME_PROPERTIES( b_frame_ptr ), "hide" );
+
+ // Process the transition
+ *frame = mlt_transition_process( this, a_frame_ptr, b_frame_ptr );
+
+ // We need to ensure that the tractor doesn't consider this frame for output
+ if ( *frame == a_frame_ptr )
+ b_hide |= type;
+ else
+ a_hide |= type;
+
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES( a_frame_ptr ), "hide", a_hide );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES( b_frame_ptr ), "hide", b_hide );
}
- return 0;
- }
- if ( index == b_track )
- {
- // Pass on the 'b frame' and remember that we've done it
- *frame = this->b_frame;
- this->b_held = 0;
- return 0;
}
+
+ // Obtain the frame from the cache or the producer we're attached to
+ if ( index >= a_track && index <= b_track )
+ *frame = this->frames[ index ];
else
- {
- // Pass through
- return mlt_service_get_frame( this->producer, frame, index );
- }
+ error = mlt_service_get_frame( this->producer, frame, index );
+
+ // Determine if that was the last track
+ this->held = !mlt_properties_get_int( MLT_FRAME_PROPERTIES( *frame ), "last_track" );
+
+ return error;
}
/** Close the transition.
else
{
mlt_service_close( &this->parent );
+ free( this->frames );
free( this );
}
}