Smooth playback issues with mp4live

Hi,

I am developing a new video source for a custom piece of
hardware (non
v4l) that I want to use within the mp4live framework.

I have it mostly working, except when I finally view it the
playback is
a little slow and choppy.

Now if I use the "preview source" functionality of
the mp4live gui I see
a nice smooth video. Does this mean that my video source is
working
properly and perhaps it is the encoding library? Or does
this the
preview source not offer a robust test in fixing this?

I'm just not quite sure where to start debugging this
because it looks
to me that my source is working fine, but the end result
isn't good.

Thanks,
Jon


------------------------------------------------------------
-------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and
take
control of your XML. No limits. Just data. Click to get it
now.
http://sourcefor
ge.net/powerbar/db2/
_______________________________________________
Mpeg4ip-discussion mailing list
Mpeg4ip-discussionlists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mpeg
4ip-discussion

The source preview has really nothing to do with
smooth playback.

The preview is a "get a frame, show a frame",
with
no timing information at all.

The playback is based on timestamps; most likely,
the timestamps that you are passing are not
correct, or are inconsistent (that's why we
want you to use the v4l2 interface; the frames
are timestamped upon capture).

I would start with there; making sure that
the timestamps are consistent.

You can add a debug_message in
CVideoEncoder::ProcessVidoeYUVFrame.

Bill

Jon Dufresne wrote:
> Hi,
> 
> I am developing a new video source for a custom piece
of hardware (non
> v4l) that I want to use within the mp4live framework.
> 
> I have it mostly working, except when I finally view it
the playback is
> a little slow and choppy.
> 
> Now if I use the "preview source"
functionality of the mp4live gui I see
> a nice smooth video. Does this mean that my video
source is working
> properly and perhaps it is the encoding library? Or
does this the
> preview source not offer a robust test in fixing this?
> 
> I'm just not quite sure where to start debugging this
because it looks
> to me that my source is working fine, but the end
result isn't good.
> 
> Thanks,
> Jon
> 
> 
>
------------------------------------------------------------
-------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2
express and take
> control of your XML. No limits. Just data. Click to get
it now.
> http://sourcefor
ge.net/powerbar/db2/
> _______________________________________________
> Mpeg4ip-discussion mailing list
> Mpeg4ip-discussionlists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/mpeg
4ip-discussion
> 

------------------------------------------------------------
-------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and
take
control of your XML. No limits. Just data. Click to get it
now.
http://sourcefor
ge.net/powerbar/db2/
_______________________________________________
Mpeg4ip-discussion mailing list
Mpeg4ip-discussionlists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mpeg
4ip-discussion

> The playback is based on timestamps; most likely,
> the timestamps that you are passing are not
> correct, or are inconsistent (that's why we
> want you to use the v4l2 interface; the frames
> are timestamped upon capture).

Currently I get the timestamp from the mpeg2ts that is
transferred to
user space from the hardware.

I did notice something. If I put a delay equal to the frame
time period
between ForwardFrame() calls inside the video source, the
playback is
much smoother (but not perfect). This hints to me that the
Timestamp is
less important then you let on because it seems mp4live will
do
something different with the video frame based on time of
arrival and
not simply the Timestamp it is given. Does this make sense?
I suppose I
could write a function to delay up to the actual Timestamp
of the frame,
but I'm worried that will be inconsistent due to
unpredictability of
software threads.

Thanks for the help and any other ideas you may have,
Jon


------------------------------------------------------------
-------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and
take
control of your XML. No limits. Just data. Click to get it
now.
http://sourcefor
ge.net/powerbar/db2/
_______________________________________________
Mpeg4ip-discussion mailing list
Mpeg4ip-discussionlists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mpeg
4ip-discussion