On 6/28/07, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> Yes and no. It is perfectly feasible to create a video > from a stream of photographs, with each picture > translating to a single frame of video, but you can't > just append it to something that's already been encoded. > The way that video is encoded, you can't just slap an > extra frame onto the end. I don't get it, why is the encoding process so different? Isn't like instead of encoding at full speed, you "freeze" the process after each frame? If a buffer of some pictures is needed I can keep it.
I guess Ivan meant a different thing than you do, something like appending a new frame to an already encoded video WITHOUT reencode the whole thing. That's feasible (with some tricks, in transcode) only if used codec compress intraframe only (MJPEG come to mind). But IIUC you just want to freeze te encoder after each frame encoded, I am correct? If so, it should be already feasible with socket protocol and some shell script magic. Maybe (I've little practice on this field, notably socket tricks) transcode lacks some features like notification of a frame just encoded, I just don't know; anyway, going socket is definitively the way to go IMHO (and one of cleanest way that I see).
No there won't be much difference between frames and I am not going to use all i-frames; I'd like to encode the video in a format like flashvideo or similar.
Just take in account that transcode doesn't yet support flashvideo nor some `exotic' codecs (patches welcome :) )
Here you can find some examples encoded after the capture was finished: http://www.manoweb.com/alesan/timelapse/
That's quite nice :) +++ Casomai servisse, per la risoluzione di questo problema puoi anche contattarmi in privato e in italiano (visto che la lingua delle ML e` l'inglese :) ) +++ Bests, -- Francesco Romani
