[mythtv] mythtv-vid branch needs testing

Daniel Kristjansson danielk at cuymedia.net
Wed Aug 1 19:28:31 UTC 2007


Aside from a memory leak it looks like most of the OSX issues have
been addressed. A few of us have been using this branch for a while,
but it would be good to get some people to test this before we merge
it into the trunk.

Instead of only being able to pick one video decoder, one video
rendering method, one OSD rendering method and one deinterlacing
method at a time, the mythtv-vid branch allows you to configure
different methods depending on the resolution of the video. 
You can also chose the fallback deinterlacer to use if the first
once can't be used; useful if you want to fallback to kernel
deinterlacing or linear blend when bob deinterlacing fails.

There are also some additional rendering methods: For example, OpenGL
rendering, which uses OpenGL 2.0 instead of XVideo for color conversion,
scaling and compositing. Another one is XvMC-OpenGL rendering which
uses an nVidia extension to allow XvMC decoded frames to be rendered
using OpenGL with OpenGL composited OSD. Both methods require greater
CPU and GPU resources, but can result in a better looking video. The
XvMC-OpenGL rendering method will not be compiled in unless you link
directly to the XvMCNVIDIA library using the --xvmc-lib ./configure
parameter.

Three video profiles groups are defined by default "CPU++", "CPU+",
"CPU--". The first "CPU++" is the default for all frontends and 
just uses ffmpeg and the old default video rendering method. The
second "CPU+" uses ffmpeg and XVideo for SDTV resolutions and then
degrades to the lower output quality XvMC and libmpeg2 decoders for
higher resolutions. Finally, "CPU--" tries to use hardware decoding
whenever possible and falls back to libmpeg2 with XVideo. The profile
groups are global for all your for all your frontends, but you can
configure which one to use on any particular frontend and create new
ones if you wish. The first matching rule which is usable with your
hardware is used, so you can have ivtv configured even if you don't
have the hardware. If no matching rule is found MythTV will pick
a decoder and rendering methods which will work, but of course these
may not be ideal for your particular hardware.

-- Daniel



More information about the mythtv-dev mailing list