[FFmpeg-devel] Would like to work on lavf-based concatenation tool for GSoC qualification

[Robert Swain]

On 31/3/09 16:10, Kostya wrote:
> On Tue, Mar 31, 2009 at 03:26:37PM +0100, Robert Swain wrote:
>> On 31/3/09 15:00, Geza Kovacs wrote:
>>> I'd like to know, by "concatenation" I assume this means "without
>>> re-encoding"; therefore I assume I'm not supposed to handle situations
>>> when multiple files are encoded with different framerate settings or
>>> different codecs (or do I instead transcode them into an intermediate
>>> format in these situations?).
>> Concatenate means to link to together. I'm not sure what the plans are
>> for handling differences in frame rates or resolutions. Differing source
>> formats shouldn't matter for transcoding purposes as long as on entry to
>> the video encoder the resolution, frame rate and pixel format as all
>> appropriate and similarly the number of channels, sample rate and sample
>> format for the audio encoder, there shouldn't be any major problem.
>>
>> I expect Baptiste will elaborate on this. I think he had plans to allow
>> specification of command line options per input so that one could affect
>> the different inputs to make them concatenate-able.
>
> That's rather for full GSoC task (enhancing FFmpeg). Small task is writing
> simple tool for at least reading data in the same format and write it to
> output. Enhancing it would be good and ease GSoC task.

Oh, I didn't notice that it was down as a small task for qualification 
too. Yes, something that rejects any streams with significant 
differences and otherwise works with stream copy would be good.

What should happen with timestamps in these cases? I guess they should 
follow on from where the previous batch finished.

Regards,
Rob