Proteus Autodetect Parameter scan

Is it possible to use cli to use the model prap-v3 just to spew the suggested frame parameters to a text file?, i know i can just click and check the log files for them, but i want to do some graph on the auto detect parameters, without up scaling with Proteus v4, so that i can format them to a csv file and work with them and a timeline in a graphing program.

Can you get them form the logs? Maybe I can make a little converter script to pump them into a csv.

As cool as this idea is, I can only see it being a way to visualize what cannot be changed or manipulated.

each time you select, autodetect, and it gives you values, those values are logged, but making a script to click all frames is annoying

probably, but i could make graphs of a before and after processing, and trying to put in bars how much compression artifacts are present before and after

Looking at the logs, it seems the model would be tvai_pe=model=prap-3, although i havent tested, i was just looking at a log, i wished the topaz video ai cli wiki would have more documentation

i ran this command ffmpeg.exe -nostdin -hide_banner -y -i “test.mp4” -vf tvai_pe=model=prap-3 -f null - and all i’m getting is “swscaler no accelerated colorspace conversion found from yuv420p to rgb4813e”

This actually works!, i just had an issue with setting variables, i can now get a full scan of the video, i may start doing a script to pass it to a cvs file

I’m curious to see what you end up creating.

me too

tbh i was thinking of doing this to do a prescan of the video, probably separate it by scenes, and then run proteus with a wider parameter scan than 20 frames for more accurate results

1 Like