On Sun, Jan 12, 2014 at 8:21 PM, Hendrik Leppkes <[email protected]> wrote: > On Sun, Jan 12, 2014 at 8:02 PM, Carl Eugen Hoyos <[email protected]> wrote: >> Hendrik Leppkes <h.leppkes@...> writes: >> >>> image2 requires to directly access the files itself. >> >> Thank you for correcting me! >> >>> There is also a image2pipe, which can read the data from a avio >>> context, but in my experience it also didn't work properly because its >>> designed for a "pipe" scenario where it doesn't know the file size, >>> and therefor reads kinda blindly. >> >> Could you elaborate a little on the shortcomings of image2pipe? >> I seem to remember that it is used often. >> Or do you just mean that too many parsers are missing? > > Yeah it only works on img formats which have a parser to re-assemble > full images, so its use is a bit limited. > On the other hand, you can actually use it to pipe several images in > one command into it (if you have a parser), so both use-cases can be > useful i suppose.
Could you elaborate a bit what would be needed to make this work, i.e. demuxing single images (not a series) like jpg, png, dpx and probably others covered by the image2 demuxer with a custom AVIOContext that has a size and in my case can also seek but without access to the file? _______________________________________________ Libav-user mailing list [email protected] http://ffmpeg.org/mailman/listinfo/libav-user
