[FFmpeg-user] Convert a series of SGI images to video

[Paul B Mahol]

On 3/2/13, Werner Robitza <werner.robitza at gmail.com> wrote:
>> > The question is, why shouldn't it?
>> > cat *.jpg | ffmpeg -f image2pipe -c:v mjpeg -i - works as far as I
>> > know,
>> > and is used here as well:
>> >
>> https://ffmpeg.org/trac/ffmpeg/wiki/Create%20a%20video%20slideshow%20from%20images
>>
>> That works for jpeg, png, bmp, ppm, pgm, pbm, pam but can not for sgi
>> as there is no
>> sgi parser.
>
>
> Interesting, thanks.
>
> What do you mean by "SGI parser", specifically?
> Where is the difference between that kind of parser and a demuxer (which
> FFmpeg does have for SGI)?

FFmpeg does have SGI decoder. SGI parser would split stream (in this
case concatenated SGI files) into packets that would be feed to SGI
decoder. But having parser is not strictily necessary, image2pipe
demuxer could be changed to use returned value of (SGI) decoder to
know where next image begins, but once again that would be hack as you
would not know ahead from demuxer how big packets should be feed to
decoder.
There is no point to have SGI demuxer as such one would actually do
parser job, in your case demuxer is image2pipe.