[FFmpeg-devel] Testing hardware H264 encoder (probably OT)

[Michael Niedermayer]

On Sat, Apr 07, 2012 at 09:35:41AM +0200, Andrea Marson - DAVE Simply Embedded wrote:
> Fist of all I would apologize if this post is not strictly related
> to the topic of this mailing list. I hope that some video experts
> will kindly give me some hints or useful links that allow me to
> continue my research elsewhere.
> 
> I'm performing some tests on hardware H264 encoder integrated in a
> system-on-chip by Texas Instruments. The first thing I have to do is
> to prove it is able to encode a 1080p60 video without any frame
> drop.
> To do this test, I built with ffmpeg and ImageMagick a 1080p60 video
> where every frame has been "tagged" by a text showing a progressive
> counter starting from 1 at the first frame. This allows to identify
> each single frame in the video. I feed the encoder with this video
> and I store the output stream on a file. By inspecting the output
> file on a frame-by-frame basis I'm able to verify how the encoder
> processed the input video.
> With current configuration encoder drops around 600 frames out of
> 2800. I would like to change one of encoder parameters (profile,
> level, bitrate etc.) at a time in order to find the proper
> combination that leads to no frame drop. Since there are tons of
> parameters, I would not change these randomly so I'd like to
> understand which parameters I should look at first. Secondly I'd
> like to know if there are different test suites that are more
> efficient than my rough test procedure.
> 
> Any help will be really appreciated.

If you want to get encoder X to do Y best would be to reads encoders
X manual or if its something you bought, licensed or so. There may
be a a support hotline.
Not sure why you think we would know how to configure some random
hardware encoder, is there some relation to ffmpeg ? is that hardware
decoder using ffmpeg ?


[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20120407/585332ce/attachment.asc>