[theora] Fwd: ffmpeg2theora 0.24 regression: accelerated video output (converted from h264)
disturbed1976 at gmail.com
Thu Jul 16 07:38:57 PDT 2009
On Thu, Jul 16, 2009 at 9:22 AM, <xiphmont at xiph.org> wrote:
>>> This is an all too common error with ffmpeg not correctly identifying
>>> the input fps.
>> Hi Keith!
>> Nice shot! Adding "--inputfps 25 -F 25" fixes my issue.
> It has become a common tactic for containers to always declare 1000fps
> or some other absurdly large value, and let the PTS on the video/audio
> packets dictate the actual sync. For a fixed-framerate codec, you can
> just guess the actual fps figure and it will be reasonably obvious
> when you're right. For truly VFR codecs (like wmv), you may need
> fancier options to deduce and enfore a framerate.
The wmv issue with 1000 fps which is just wmv being a faulty container.
ffmpeg has issues with mpeg program and transport streams in PAL
content (25fps) is reported as 50FPS, same issue with telecined NTSC
content (23.976 telecined to 29.976 fps) is sometimes reported as
47.952 - discussed on this list a few weeks back. NTSC content at
29.970 fps is reported as 59.94. The pattern is ffmpeg reports the
frame rate as doubled. This is actually a long time bug I've came
across for roughly three years now. It's to the point where it's
embedded in my brain to always specify the frame rate when ever I need
to use ffmpeg.
In the op's example, ffmpeg reports the frame rate as being 50 tbc.
For NTSC material it looks like this -
Stream #0.0[0x1e0]: Video: mpeg2video, yuv420p, 720x480 [PAR 8:9 DAR
4:3], 9608 kb/s, 29.97 tbr, 90k tbn, 59.94 tbc
or with another stream -
Stream #0.0[0x1e0]: Video: mpeg2video, yuv420p, 720x480 [PAR 32:27 DAR
16:9], 9800 kb/s, 59.94 tbr, 90k tbn, 59.94 tbc
More information about the theora