[FFmpeg-user] Process for determining compression settings

Ian Pilcher arequipeno at gmail.com
Mon May 2 19:47:23 EEST 2022

This is very much a "meta" question, rather than something specific
ffmpeg, but it feels like there's so much discussion of video
settings here, that it's a reasonable subject.

The background to my question is that I am using Video Enhance AI[1] to
upscale and cleanup a number of old videos.  I've set VEAI to write the
results as a sequence on PNG images, and I now intend to use ffmpeg to
create compressed video streams (H.264 or H.265) and mux the audio back
in to a final MKV container.

My challenge is to determine my compression settings in any reasonable
sort of time, without driving myself insane.  Most of the guidance I've
been able to find seems to boil down to "try some stuff and look at the
results," which really isn't practical when one considers the sheer
number of different variables, the time required to compress even a
short video sample, the time required to watch various different
versions of the sample, and (most importantly) the inability of human
vision and memory to make any reasonable comparison between subtly
different video clips watched several minutes apart.

So I'm curious how others go about this.  Do you have a particular
workflow that you use to determine good compression settings for a
video?  Are there any video quality metrics that you find to be useful
in such a scenario (e.g. VMAF or others)?  Something else?

I can't be the only person to struggle with this, can I?


[1] https://www.topazlabs.com/video-enhance-ai
Google                                      Where SkyNet meets Idiocracy

More information about the ffmpeg-user mailing list