Documentation for ffmpeg?

Hi all, I’m playing with the beta version on Linux, for fun I was able to get it dockerized and I have it running on my local k3s cluster.

I’m playing with different settings, where can I get a full list of available models? I have a default ffmpeg filter string of tvai_up=model=prob-3:scale=2:preblur=-0.6:noise=0:details=1:halo=0.03:blur=1:compression=0:estimate=20:blend=0.8:device=0:vram=1:instances=1, but I don’t know exactly what can customize.

I know prob-3 is the model, scale is self explanatory, but I’m curious how to accomplish things here similar to in the UI. Maybe I just can’t find the documentation and I just need to go RTFM.

For example - how do I scale to a specific resolution? How can I up the framerate using interpolation? How can I do both? Maybe add in stabilization? How do I deinterlace video? Maybe none of these are available right now?

Thanks! Having a ton of fun!


This page in our documentation for Video AI is a great starting point for the CLI, and includes a full list of the “short names” for each AI model.

Another way to quickly build a command for the CLI is to select settings in the Video AI GUI and then use Control+Shift+E to reveal the CLI equivalent for the enabled models/parameters.


Thank you! I had the short names from a random script but had no idea what each one related to it’s actual context. Also had no idea about the Ctrl+Shift+E, that’s an amazing tool, going to go try that now!

That is just a start. It has no details like what are all the parameters for Proteus, and what are the value limits.

Adding syntax for the tvai filter’s command line help…

$ ffmpeg -hide_banner -h filter=tvai_up

Filter tvai_up
  Apply Topaz Video AI upscale models, parameters will only be applied to appropriate models
       #0: default (video)
       #0: default (video)
tvai_up AVOptions:
   model             <string>     ..FV....... Model short name (default "amq-13")
   scale             <int>        ..FV....... Output scale (from 0 to 4) (default 1)
   w                 <int>        ..FV....... Estimate scale based on output width (from 0 to 100000) (default 0)
   h                 <int>        ..FV....... Estimate scale based on output height (from 0 to 100000) (default 0)
   device            <int>        ..FV....... Device index (Auto: -2, CPU: -1, GPU0: 0, ...) (from -2 to 8) (default -2)
   instances         <int>        ..FV....... Number of extra model instances to use on device (from 0 to 3) (default 0)
   download          <int>        ..FV....... Enable model downloading (from 0 to 1) (default 1)
   vram              <double>     ..FV....... Max memory usage (from 0.1 to 1) (default 1)



I have a problem with Command Line; I copied a ffmpeg command line from Topaz Video AI, then I used this command line to handle a video in ‘cmd’, I got two results, one from the software and another from the command line, they two are different, have you ever encountered this kind of problem?

This is the command line I copied:
’ ffmpeg “-hide_banner” “-i” “input.ts” “-sws_flags” “spline+accurate_rnd+full_chroma_int” “-color_trc” “6” “-colorspace” “6” “-color_primaries” “6” “-filter_complex” “tvai_up=model=nyx-2:scale=2:w=3840:h=2160:preblur=0:noise=0:details=0:halo=0:blur=0:compression=0:estimate=8:device=0:vram=1:instances=0,scale=w=3840:h=2160:flags=lanczos:threads=0,colorspace=ispace=5:space=6:primaries=6:trc=6” “-c:v” “h264_nvenc” “-profile:v” “high” “-pix_fmt” “yuv420p” “-g” “30” “-preset” “p7” “-tune” “hq” “-rc” “constqp” “-qp” “25” “-rc-lookahead” “20” “-spatial_aq” “1” “-aq-strength” “15” “-b:v” “0” “-map” “0:a” “-map_metadata:s:a:0” “0:s:i:2” “-c:a” “copy” “-bsf:a:0” “aac_adtstoasc” “-map_metadata” “0” “-map_metadata:s:v” “0:s:v” “-movflags” “frag_keyframe+empty_moov+delay_moov+use_metadata_tags+write_colr” “-metadata” “videoai=Enhanced using nyx-2; mode: auto; reduce noise at 0; reduce large grain at 0; revert compression at 0; dehalo at 0; and anti-alias/deblur at 0. Changed resolution to 3840x2160” “output.mp4” ’