[FFmpeg-devel] [PATCH] [RFC] PNG: decode textual data

[Reimar Döffinger]

On Sun, Mar 20, 2011 at 11:58:28AM +0100, Nicolas George wrote:
> The attached patch implements decoding of textual data inside PNG images.
> 
> It is mostly useless right now because there is no API to return the data to
> the user: the metadata API would be the preferred way, but it is in
> libavformat and this not usable from libavcodec.
> 
> A possible solution is:
> 
> 1. move metadata.* from libavformat to libavcore... ahem... libavutil;
> 
> 2. add an AVMetadata structure to AVFrame.
> 
> Any advice?

I'd be in favour of a more general "per-frame userdata" since that is
also needed for CC subtitling with DVB/ATSC.
It makes sense to do this in the decoder and not the demuxer because
that data needs to be reordered (and also it should be higher performance
I think).
I guess AVMetadata might work, but it is rather badly suited for binary
data for now at least.