Decoded data in libav* is stored in an AVFrame structure. In the case of video,
this is a complete frame of video. You create one of those, and fill out its
data, and encode it with avcodec_encode_video. For the purposes of encoding an
image, you're really just creating a "video" with one frame. (Similarly, for
decoding, you treat the jpeg file as a "video" with one frame.)
The output-example.c example shows how to create the AVFrame and encode it.
There is another sample app in libavcodec/api-example.c in the source
distribution to look at too. If you need to convert between pixel formats (e.g.
rgb to yuv), see the sws_scale() function. (The frame given to the codec to
encode has to have a pixel format compatible with the codec. I use
PIX_FMT_YUVJ420P for jpeg. Those constants are in libavutil/pixfmt.h.)
Study the sample code, and I would also generate doxygen docs of the libav*
source to help you navigate around and familiarize yourself with the APIs and
how they work.
Andy
On 7/22/2010 11:15 AM, Denis Gottardello wrote:
Alle giovedì 22 luglio 2010, Michael Chisholm ha scritto:
You ought to be able to decode and encode jpegs with libav's builtin jpeg
codec. I've done both before with the "mjpeg" codec. I didn't need
libjpeg. It's not C++ though. You'd have to either use C from within a C++
app, or find some other library. From the looks of things, libjpeg isn't
C++ either...
Ignoring the language which is the rigth way to use the ffmpeg libraries to
transalte a just in memory rgb (or yuv) image to jpeg image?
_______________________________________________
libav-user mailing list
[email protected]
https://lists.mplayerhq.hu/mailman/listinfo/libav-user
