Why does `inverse_telecine.md` use yadif?
I looked at the inverse_telecine.md file, and it recommends using YADIF.
I don't think you should need to deinterlace once you convert from 29.970fps to 23.976fps because if the original source is 23.976fps, it should be a 1:1 conversion with a single extra frame which is why you noted the "decimate" command right?
Also, your command doesn't take into account the fact that ffmpeg reduces quality significantly at its default settings. It even converts my audio track to OGG for some reason.
Hi @Sawtaytoes! I'll be honest, I wrote these scripts a loooong time ago now, and it was just fun experimentation mostly. I'm surprised that anyone has looked at this repo in the last five years :laughing:
I don't think you should need to deinterlace once you convert from 29.970fps to 23.976fps because if the original source is 23.976fps, it should be a 1:1 conversion with a single extra frame which is why you noted the "decimate" command right?
The need for deinterlacing is because of the blended frames that the 3:2 processes introduces. The Wikipedia article explains it well - here is an image from there illustrating the issue:
I guess that your source simply repeats frames without blending? If so, then yeah, you could just decimate without deinterlacing.
Also, your command doesn't take into account the fact that ffmpeg reduces quality significantly at its default settings. It even converts my audio track to OGG for some reason.
It's been a long while since I've looked at any of this stuff, but outside of toy scripts like this, I always encoded using pretty high settings, e.g. -preset veryslow and a good crf. That would be my recommendation anyway.
I haven't come across the OGG issue - how peculiar!