Re: [FFmpeg-devel] Help for patch: Sending last full, non-corrupt frame to output

2015-09-03 Thread Carl Eugen Hoyos
Chris Wiggins chriswiggins.co.nz> writes: > Should I just disable threads when I'm testing to > see if this alleviates the problem? I would expect that the output frame looks different with single-threaded decoding. In any case, a real-world h264 stream may help improving the issue. Carl Eug

Re: [FFmpeg-devel] Help for patch: Sending last full, non-corrupt frame to output

2015-09-02 Thread Chris Wiggins
Carl, > Did you save the (h264) source for this output? > (Could it be threads-related?) I haven't saved the source but I'll try and replicate the problem tomorrow by dropping frames at iptables and get back to you. The cameras that were causing issues aren't online at the moment. Should I ju

Re: [FFmpeg-devel] Help for patch: Sending last full, non-corrupt frame to output

2015-09-02 Thread Kieran Kunhya
On 2 September 2015 at 08:30, Carl Eugen Hoyos wrote: > Chris Wiggins chriswiggins.co.nz> writes: > >> however sometimes it washes the frame out: >> http://i.imgur.com/o5YttPY.png > > Did you save the (h264) source for this output? > (Could it be threads-related?) No, its packet loss and/or reor

Re: [FFmpeg-devel] Help for patch: Sending last full, non-corrupt frame to output

2015-09-02 Thread Carl Eugen Hoyos
Kieran Kunhya obe.tv> writes: > >> http://i.imgur.com/o5YttPY.png > > > > Did you save the (h264) source for this output? > > (Could it be threads-related?) > > No, its packet loss and/or reordering. Can you provide a sample? I would expect the issue to be threads-related. Carl Eugen

Re: [FFmpeg-devel] Help for patch: Sending last full, non-corrupt frame to output

2015-09-02 Thread Carl Eugen Hoyos
Chris Wiggins chriswiggins.co.nz> writes: > however sometimes it washes the frame out: > http://i.imgur.com/o5YttPY.png Did you save the (h264) source for this output? (Could it be threads-related?) Carl Eugen ___ ffmpeg-devel mailing list ffmpeg-de