[FFmpeg-devel] [PATCH V2 0/3] Add libsvt HEVC encoder wrapper

mypopy at gmail.com mypopy at gmail.com
Wed Dec 5 03:25:24 EET 2018

On Wed, Dec 5, 2018 at 7:42 AM Mark Thompson <sw at jkqxz.net> wrote:
> On 04/12/2018 15:23, Moritz Barsnick wrote:
> > On Tue, Dec 04, 2018 at 22:25:29 +0800, Jun Zhao wrote:
> >> This wrapper work with SVT-HEVC new_api branch, more information can get
> >> from https://github.com/intel/SVT-HEVC/blob/new_api/ffmpeg_plugin/.
> >
> > Is this API stable? (I have heard voices that the external library
> > wrapper shouldn't aim at a moving target.)
> >
> > If so, can the API/version be detected (e.g. vs. the one on master), or
> > will the compile against a build from master just fail? (Or is this
> > branch used only because ffmpeg_plugin exists only on this branch?)
> This looks like a big problem currently.  The codebase  appears more like a prototype rather than finished code - the headers are very dirty, and it's leaking all sorts of weird symbols into the global namespace.  Given that, it seems unlikely that the current form of the headers and library will survive to release, and therefore we should probably hold off for now.
As far as know, SVT-HEVC come from eBrisk video, more information can
be found by: https://www.intel.com/content/www/us/en/processors/xeon/scalable/software-solutions/ebrisk-comms-video.html,
I don't think the SVT-HEVC codebase just as the prototype, I know
SVT-HEVC codebase need to refine and refactoring, and ä»–his is exactly
what these people are doing on new_api branch.

BTW: If we declined to include the wrapper as part of FFmpeg mainline,
this means SVT-HEVC need to maintain the FFmpeg wrapper patch set for
this, I don't think this is a good idea if some guys want to try
SVT-HEVC encoder from FFmpeg.
> Relatedly, are there any results indicating how good this encoder actually is, especially as compared to libx265 as the main existing software H.265 encoder?  FFmpeg previously declined to include a wrapper for the Turing H.265 encoder, due to lack of any clear use case for it at the time (that might have changed by now, I don't know).
maybe can answer part of the above question. Tks.
> Thanks,
> - Mark

More information about the ffmpeg-devel mailing list