[FFmpeg-user] size and bitrate issues with nvidia accelerated transcoding

Sreenath BH bhsreenath at gmail.com
Tue Mar 15 10:40:58 CET 2016

On 3/9/16, Andy Furniss <adf.lists at gmail.com> wrote:
> Sreenath BH wrote:
>> On 3/8/16, Moritz Barsnick <barsnick at gmx.net> wrote:
>>> If you want the same profiles and levels, you will need to request them
>>> explicitly.
>>> The same thing goes for the quality: libx264 defaults to crf=23, I
>>> don't know about nvenc. Furthermore, encoders can trade off quality (or
>>> size) for conversion speed. So if you want the same size, you have to
>>> specify "-b:v" explicitly. If you want the same quality, I guess you
>>> may have to experiment with the settings available to find something
>>> comparable.
>> Thanks for your comments. I understand that two codecs even with same
>> presets and other similar settings can produce differing output.
>> I wanted to know if there was a way to achieve the results I wanted
>> with the codec we prefer to use (nvidia) because it have significant
>> speed increase.
> Did you try what Moritz wrote above?
> Also see nvenc section of
> https://trac.ffmpeg.org/wiki/HWAccelIntro
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-user


I have tried several of the options provided by the nvidia encoder and
in all cases, the only virtue of the nvidia encoder seems to be speed.
As soon as I start limiting the target bitrate (using -b:v option),
the video quality suffers in fast changing areas of the picture.

We will continue to use the nvidia for some time to come and hope to
see an updated version that can work better at reduced bitrates.


More information about the ffmpeg-user mailing list