[FFmpeg-devel] [RFC} extend pixdesc ?
stefano.sabatini-lala at poste.it
Sun Mar 20 01:07:58 CET 2011
On date Saturday 2011-03-19 19:41:07 +0000, Mark Himsley encoded:
> Dear Developers,
> It was previously noted in the thread "[FFmpeg-devel] libavfilter:
> extending overlay filter" that each filter is doing its own
> investigation to discover the layout of RBGA style formats, to discover
> if the format has alpha, to discover the level of black.
> Would it be appropriate to extend AVPixFmtDescriptor to contain a new
> struct holding that information, and add that information to
> av_pix_fmt_descriptors in pixdesc.c?
This idea pops up from time to time. The problem is that the pixel
format, and consequently the pixel descriptor, should only express the
layout of the pixel components, while chroma and colorspace
information should be placed somewhere else.
Unfortunately until we properly implement this, we need to rely on
various hacks for determining chroma/colorspace information from the
pixel format, but we are trying to avoid to pollute the pixdesc struct
with information which semantically doesn't belong there.
But I agree in the meaningwhile we may rely at least on shared
internal routines for extracting this information from the pixel
FFmpeg = Frenzy and Fast Mastering Prodigious Embarassing Gymnast
More information about the ffmpeg-devel