Some older videos might be converted from VHS or recorded from TV signals or created by camcorders, etc…
Those exhibit specific artifacts, for example:
- heavy chromatic aberration
- interlacing artifacts encoded into progressive video
- artifacts from heavy use VHS/taped video
I understand that this might be a somewhat niche use-case which also might be difficult to train AI models for due to the lack of “clean” sources.
Just an idea…