How to seek in frames and schedule a schedule a single shot clock id correctly

[Marcus Nascimento]

An important detail, I'm targetting gstreamer 0.10.3 since that's what's
available in my embedded environment.
If it got better in newer versions, I may try to cross compile it.

Thanks again.


On Thu, Nov 7, 2013 at 3:21 PM, Marcus Nascimento <marcus.cps at gmail.com>wrote:

> I'm trying to create a media player that must supports the following
> features:
>
> - Set Start and/or End position using TIME (timestamp);
> - Set Start and/or End position using SAMPLE (for audio only);
> - Set Start and/or End position using FRAME (for video only);
> - Schedule a notification in a given TIME (timestamp);
> - Schedule a notification in a given SAMPLE (for audio only);
> - Schedule a notification in a given FRAME (for video only).
>
>
> It is easy to do things when the Start position where set using the same
> unit as the notification (both using TIME or SAMPLE or FRAME).
> I managed to find a way to convert SAMPLE# to TIME and vice versa (the
> opposit).
> Notice the END position doesn't really matter.
>
> But when I try to set the Start position in FRAMEs and schedule a
> notification in TIME, it gets complicated, since I can't get the base time
> to schedule the notification.
>
> I couldn't also make it work when I set the Start position in TIME and
> schedule a notification in FRAMEs. I can't find which is the frame
> corresponding to the Start position.
>
> Does anyone know whether it is possible to do that?
>
> Here is how I do things:
>
> This is how I set the END position:
>
> gst_element_seek(pipeline, 1.0, endPosFormat,
> (GstSeekFlags)(GST_SEEK_FLAG_FLUSH |
> GST_SEEK_FLAG_ACCURATE), GST_SEEK_TYPE_NONE, 0, GST_SEEK_TYPE_SET,
> endPosition);
>
>
> This is how I set the START position:
>
> gst_element_seek(pipeline, 1.0, startPosFormat,
> (GstSeekFlags)(GST_SEEK_FLAG_FLUSH | GST_SEEK_FLAG_ACCURATE),
> GST_SEEK_TYPE_SET, startPosition, GST_SEEK_TYPE_NONE, 0);
>
>
> startPostFormat and endPosFormat = FST_FORMAT_TIME for TIME.
> startPostFormat and endPosFormat = GST_FORMAT_DEFAULT for SAMPLEs or
> FRAMEs.
>
>
> 1) When setting the Start position in TIME, I do this:
>
> - Set the start time using the seek instruction above.
> - gst_element_set_start_time(pipeline, GST_CLOCK_TIME_NONE);
> - gst_element_set_base_time(pipeline, startPosition);
>
>
> 2) When setting the Start position in SAMPLEs, I do this:
>
> - Set the start sample using the seek instruction above
> - Convert the startPosition from SAMPLE to TIME using the RATE (got from
> the caps from the sink pad of the audio-sink element).
> - gst_element_set_start_time(pipeline, GST_CLOCK_TIME_NONE);
> - gst_element_set_base_time(pipeline, startPosition);
>
>
> 3) When setting the Start position in FRAMEs, I do this:
>
> - Set the start frame using the seek instruction above.
> - I can't find a corresponding TIME for this FRAME, since I can't find a
> FRAME RATE nor can guarantee it would be constant. Is it possible?
>
>
> 4) When scheduling a TIME notification, I do this:
>
> - GstClock *clock = gst_pipeline_get_clock(GST_PIPELINE(pipeline));
> - GstClockID *clockId = gst_clock_new_single_shot_id(clock,
> timeInNanoSecs);
> - gst_clock_id_wait_async(clockId, MyCallback, NULL);
>
>
> 5) When scheduling a SAMPLE notification, I do this:
>
> - Convert SAMPLE to TIME using the same procedure described in 2.
> - Schedule the notification using the same procedure described in 4.
>
>
> 6) When scheduling a FRAME notification, I do this:
>
> - Create a pad probe in the sink pad of the videosink element, using
> gst_pad_add_buffer_probe.
> - probeCounter = startPosition // Frame #.
> - In the probe callback, I always increment the probeCounter and check if
> it is greater or equal to target frame. If so, I call the notification
> function.
>
>
> Thanks very much,
>
> --
> Marcus Nascimento
>
>


-- 
Marcus Nascimento
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/gstreamer-devel/attachments/20131107/f95e3eac/attachment-0001.html>